Latest / iwon™ podcast with Michael J Thomas / AI Skills and Future of Work - Building a Gold Standard That Doesn't Exist (Yet) | Aatish Thakerar | iwon™ podcast Ep. 9.

Transcript

AI Skills and Future of Work - Building a Gold Standard That Doesn't Exist (Yet) | Aatish Thakerar | iwon™ podcast Ep. 9. · iwon™ podcast with Michael J Thomas

0:00
Hi,
0:00
it's
0:00
Michael
0:00
J.
0:01
Thomas
0:01
and
0:01
this
0:02
is
0:02
the
0:02
i1
0:02
podcast.
0:03
We
0:03
obviously
0:04
don't
0:04
do
0:04
polished
0:05
PR
0:05
puff
0:05
pieces,
0:06
any
0:06
of
0:06
that
0:06
stuff.
0:06
We
0:07
tell
0:07
the
0:07
real
0:07
true
0:08
story
0:08
about
0:09
what
0:09
it
0:09
takes
0:09
because
0:10
it
0:10
wasn't
0:10
overnight.
0:11
And
0:11
today
0:11
I've
0:12
got
0:12
a
0:12
really,
0:12
really
0:13
interesting
0:14
guest
0:14
today.
0:15
I'm
0:15
joined
0:15
by
0:15
Atish
0:16
Thakkarar
0:17
who's
0:17
co
0:18
founder
0:18
of
0:18
Pedagog.
0:19
And
0:20
what
0:20
they
0:21
really
0:21
want
0:21
to
0:21
do
0:21
is
0:22
obviously
0:23
make
0:23
AI
0:24
much
0:25
more
0:25
of
0:25
a
0:25
transparent
0:26
verbal
0:26
skill,
0:27
get
0:27
it
0:27
through
0:27
schools
0:28
and
0:28
really
0:29
make
0:29
it
0:29
work
0:29
for
0:30
people
0:30
in
0:30
the
0:30
future
0:31
and
0:31
be
0:31
the
0:31
gold
0:31
standard
0:32
as
0:33
his
0:33
LinkedIn
0:34
profile
0:34
says.
0:35
So
0:35
we
0:35
had
0:35
a
0:36
great
0:36
conversation
0:37
before
0:38
we
0:38
started.
0:39
Atish,
0:39
would
0:39
you
0:39
like
0:40
to
0:40
add
0:40
anything
0:40
to
0:40
what
0:41
I've
0:41
just
0:41
said
0:42
about
0:42
where
0:43
you
0:43
want
0:43
to
0:43
take
0:43
it?
0:44
I
0:44
find
0:44
you
0:44
quite
0:45
an
0:45
interesting
0:45
guy
0:46
because
0:47
you
0:48
have
0:48
a
0:49
humanities
0:49
background.
0:51
You
0:51
were
0:51
in
0:51
the
0:51
British
0:52
army,
0:52
you,
0:52
you
0:53
were
0:53
chief
0:54
engineering
0:54
officer
0:55
at
0:55
climatx
0:56
and
0:57
now
0:57
you
0:58
took
0:58
the
0:58
brave
0:58
step
0:59
to
0:59
launch
0:59
your,
1:00
your
1:00
own
1:00
thing
1:01
in
1:01
the
1:02
age
1:02
of
1:02
AI.
1:03
Yeah,
1:03
of
1:03
course.
1:04
So,
1:04
so
1:04
this
1:04
isn't
1:04
actually
1:05
my
1:05
first
1:05
AI
1:06
startup,
1:06
but
1:06
my,
1:07
my
1:07
first
1:08
AI
1:08
startup
1:08
came
1:09
in
1:09
the,
1:10
the
1:10
wake
1:10
of
1:10
the
1:11
great,
1:11
suppose
1:11
the
1:11
era
1:12
of
1:12
cheap
1:12
money
1:13
and
1:13
the,
1:13
the
1:14
post
1:14
Covid
1:14
era.
1:15
So
1:15
about
1:15
2020
1:17
or
1:17
so
1:18
2020,
1:20
that
1:20
sort
1:20
of
1:20
era.
1:20
And
1:20
effectively,
1:21
I
1:22
mean
1:22
to
1:22
kind
1:22
of
1:22
long
1:23
story
1:23
short,
1:24
my,
1:24
my
1:24
co
1:24
founder
1:25
at
1:25
the
1:25
time
1:25
who's
1:26
now
1:26
based
1:26
in
1:26
the
1:26
States
1:27
and
1:27
I,
1:28
we
1:28
decided
1:29
to
1:29
create
1:29
a
1:30
small
1:30
to
1:30
medium
1:31
sized
1:31
large
1:32
language
1:32
model
1:33
business.
1:33
We
1:34
were
1:34
focused
1:34
very
1:35
much
1:35
on
1:35
engineering
1:35
consultancy
1:36
companies
1:37
and
1:37
so
1:37
on.
1:38
What
1:38
we
1:38
didn't
1:39
have
1:39
necessarily
1:39
at
1:40
that
1:40
time
1:40
was
1:40
the
1:40
budget
1:41
or
1:41
the
1:42
full
1:42
know
1:42
how
1:43
to,
1:43
to
1:43
turn
1:44
this
1:44
into
1:44
a
1:44
big
1:45
sort
1:45
of
1:45
commercial
1:45
operation.
1:47
We
1:47
also
1:47
had
1:47
clients
1:48
who
1:48
for
1:48
want
1:49
of
1:49
a
1:49
better
1:49
phrase
1:50
were
1:50
really
1:51
on
1:52
just
1:53
wanting
1:55
that
1:55
technology
1:56
for
1:56
themselves.
1:56
So
1:57
it
1:57
was
1:57
a
1:57
very
1:57
sort
1:58
of,
1:58
there
1:58
was
1:58
no
1:58
sense
1:58
of
1:59
open
1:59
source
1:59
or
2:00
anything
2:00
like
2:00
that
2:01
that
2:01
you
2:01
see
2:01
in
2:01
the
2:01
world
2:02
today
2:02
and
2:02
certainly
2:03
nothing
2:03
of
2:03
the
2:03
light
2:03
coming
2:04
out
2:04
of
2:04
China.
2:05
It
2:05
was
2:05
all
2:05
very
2:05
much
2:06
focused
2:06
on
2:06
this
2:07
is
2:07
great
2:07
technology,
2:08
we
2:08
want
2:08
it
2:08
for
2:08
ourselves,
2:09
our
2:09
competitors
2:10
shouldn't
2:10
be
2:10
able
2:10
to
2:10
get
2:10
our
2:11
hands
2:11
on
2:11
it.
2:11
And
2:12
so
2:12
when
2:12
we
2:12
eventually
2:13
packaged
2:13
up
2:13
that
2:13
IP
2:14
and
2:14
sold
2:15
it
2:15
and
2:15
then
2:15
wound
2:15
down
2:16
the
2:16
company,
2:16
I
2:17
decided
2:17
that
2:17
what
2:18
I
2:18
really
2:18
needed
2:18
was
2:19
some
2:19
good
2:19
old
2:19
fashioned
2:20
engineering
2:20
experience
2:21
and,
2:21
and
2:21
this
2:21
is
2:21
to
2:22
say
2:22
engineering
2:23
experience
2:23
outside
2:24
of
2:24
consultancy.
2:25
So
2:25
what
2:26
I
2:26
really
2:26
wanted
2:27
was
2:27
take,
2:28
effectively
2:29
I
2:29
suppose
2:29
in
2:29
Peter
2:30
Thiel's
2:30
world
2:30
words,
2:31
for
2:31
better
2:31
or
2:31
worse,
2:32
to
2:32
go
2:32
from
2:32
zero
2:32
to
2:33
one.
2:33
And
2:33
so
2:33
I
2:33
joined
2:34
Climbedex
2:34
and
2:35
for
2:35
four
2:35
years
2:35
there
2:35
I
2:36
played
2:36
basically
2:36
four
2:36
different
2:37
roles.
2:37
So
2:38
I
2:38
was
2:38
a
2:38
data
2:38
engineer
2:39
which
2:39
is
2:40
more
2:40
just
2:40
a
2:40
founding
2:41
engineer
2:41
role
2:42
in
2:42
year
2:42
one.
2:42
In
2:42
year
2:43
two
2:43
I
2:43
was
2:43
an
2:43
engineering
2:44
line
2:44
manager
2:45
so
2:45
I
2:45
was
2:45
called
2:45
director
2:46
of
2:46
engineering
2:46
but
2:47
I
2:47
wasn't
2:47
really
2:47
a
2:48
director
2:48
of
2:48
engineering
2:48
and
2:49
responsibilities.
2:50
In
2:50
my
2:50
third
2:51
year
2:51
I
2:51
was
2:51
a
2:51
fully
2:52
fledged
2:52
director
2:52
of
2:52
engineering
2:53
and
2:53
in
2:53
my
2:53
fourth
2:54
and
2:54
final
2:54
year
2:54
my,
2:55
my
2:55
boss
2:56
had
2:56
actually
2:56
departed,
2:57
sort
2:57
of
2:58
departed
2:58
the
2:58
company.
2:59
And
2:59
so
2:59
I
2:59
was
2:59
an
3:00
acting
3:00
exec
3:00
as
3:01
well.
3:01
And
3:01
after
3:01
those
3:02
four
3:02
years
3:02
I
3:03
felt
3:03
as
3:03
though
3:04
I
3:04
had
3:04
learned
3:05
what
3:05
I
3:05
needed
3:05
to
3:05
what
3:06
worked,
3:06
what
3:06
didn't
3:07
work
3:08
and
3:08
what,
3:08
what
3:09
to
3:09
take
3:09
forward.
3:10
I
3:10
was
3:10
a
3:10
little
3:10
bit
3:11
tired
3:11
of
3:11
delivering
3:12
for
3:12
big
3:13
banks
3:13
and
3:13
financial
3:14
services
3:14
companies.
3:14
It's
3:15
a
3:15
bit
3:15
of
3:15
a
3:15
cliche,
3:16
but
3:16
sometimes
3:16
you've,
3:16
you've
3:17
got
3:17
to
3:17
get,
3:17
get
3:17
through
3:17
the
3:18
cliche.
3:19
I
3:19
really
3:20
wanted
3:20
to
3:20
empower
3:21
people
3:21
in
3:21
this
3:22
new
3:22
technology.
3:22
I
3:23
had
3:23
a
3:23
gut
3:23
feeling
3:24
it
3:24
was
3:24
the
3:25
right
3:25
thing
3:25
at
3:26
the
3:26
right
3:26
time.
3:27
And
3:27
then
3:28
now
3:28
we're
3:28
working
3:29
with
3:29
schools
3:30
and
3:30
educators.
3:30
But
3:31
I
3:31
think
3:31
that's
3:31
probably
3:32
the
3:32
basis
3:32
of
3:32
one
3:32
of
3:33
your
3:33
other
3:33
questions,
3:33
so
3:33
I'll
3:34
stop
3:34
there.
3:35
So
3:36
interestingly,
3:37
as
3:37
many
3:37
of
3:38
you
3:38
know
3:38
who
3:38
listen
3:38
to
3:38
the
3:39
podcast,
3:39
I
3:39
chat
3:40
to
3:40
people
3:40
before
3:41
to
3:41
get
3:41
to
3:41
know
3:41
them.
3:42
And
3:42
when
3:43
I
3:43
and
3:44
Atish
3:45
spoke
3:45
last
3:45
time,
3:46
we
3:46
discussed
3:47
the
3:49
the
3:49
fact
3:49
that
3:50
fundamentally
3:50
AI
3:51
is
3:51
moving
3:51
at
3:51
such
3:52
a
3:52
rapid
3:52
pace
3:52
but
3:52
education
3:53
isn't
3:53
keeping
3:54
up.
3:54
Go
3:54
figure.
3:55
We'll
3:55
dive
3:55
into
3:55
that
3:55
in
3:56
a
3:56
second.
3:56
But
3:56
actually
3:57
what
3:57
is
3:57
interesting
3:58
is
3:59
inevitably
4:00
when
4:00
you
4:00
were
4:00
doing
4:01
that
4:01
AI
4:01
startup
4:02
with
4:02
your
4:02
engineering
4:03
and
4:03
you
4:03
were
4:04
doing
4:04
it
4:04
for
4:04
engineering
4:04
firms
4:05
and
4:05
then
4:05
acclimate
4:06
X
4:06
for
4:06
banks
4:07
and
4:07
kind
4:07
of
4:08
learning
4:08
it
4:08
all,
4:08
would
4:09
you
4:09
recommend
4:10
that
4:11
people
4:11
having
4:12
done
4:12
what
4:12
you've
4:12
done,
4:13
that
4:13
people
4:13
go
4:13
and
4:14
try
4:14
and
4:14
start
4:14
their
4:15
own
4:15
thing
4:15
or
4:15
go
4:15
and
4:16
do
4:16
the
4:16
climatex
4:17
thing
4:17
and
4:17
go
4:18
and
4:18
work
4:19
in
4:19
a
4:19
high
4:20
growth
4:20
startup
4:20
and
4:21
kind
4:21
of
4:21
see
4:21
how
4:21
it
4:21
all
4:22
fits
4:22
together.
4:22
I'd
4:23
say
4:23
it
4:23
really
4:23
depends
4:24
on
4:24
the
4:24
person,
4:25
what
4:25
they
4:25
want
4:26
to
4:26
delve
4:26
into,
4:26
what's
4:27
hot
4:27
at
4:27
the
4:27
time.
4:28
Because
4:29
I
4:29
think
4:29
success
4:30
is
4:30
a
4:30
matter
4:31
of
4:31
timing,
4:33
distribution
4:34
and
4:34
economics.
4:34
Those
4:35
aren't
4:35
my
4:35
words,
4:35
I've
4:36
stolen
4:36
them
4:36
from
4:36
somewhere
4:36
else
4:37
and
4:37
I'll
4:37
attribute
4:37
it
4:38
shortly.
4:38
But
4:39
effectively
4:41
those
4:41
three
4:41
things
4:41
are.
4:42
Those
4:42
three
4:42
factors
4:43
are
4:43
I'd
4:43
say
4:44
the
4:44
biggest
4:44
factors
4:44
by
4:45
far.
4:45
So
4:46
it
4:46
really
4:46
depends.
4:47
For
4:47
me,
4:47
having
4:47
just
4:48
come
4:48
from
4:48
a
4:49
startup
4:49
myself
4:49
and
4:50
having
4:50
packaged
4:50
it
4:51
up,
4:51
my
4:51
overriding
4:52
feeling
4:52
was
4:53
I
4:53
got
4:53
lucky
4:54
here,
4:54
to
4:54
be
4:54
really
4:55
honest,
4:55
I
4:55
got
4:55
lucky
4:56
I
4:56
didn't
4:56
fall
4:57
on
4:57
my
4:57
face.
4:58
I'm
4:58
lucky
4:58
that
4:59
my
4:59
co
4:59
founder
5:00
was
5:00
able
5:00
to
5:01
pick
5:01
bits
5:02
up,
5:02
utilize
5:02
his
5:03
own
5:03
commercial
5:04
experience
5:05
and
5:05
so
5:05
on.
5:06
And
5:06
so
5:06
for
5:06
me,
5:07
what
5:07
I
5:07
really
5:07
wanted
5:07
to
5:07
do
5:08
was
5:08
get
5:08
in
5:08
Quite
5:08
early
5:09
and
5:09
so
5:09
climb
5:10
it
5:10
out.
5:11
That
5:11
was
5:11
getting
5:12
in
5:12
just
5:12
before
5:13
the,
5:13
the
5:13
pre
5:13
seed
5:14
was
5:14
finalized
5:15
and
5:15
kind
5:16
of
5:16
coming
5:16
out
5:16
between
5:17
series
5:17
A
5:18
and,
5:18
and
5:18
B.
5:18
And
5:19
so
5:19
that
5:19
was
5:19
four
5:19
years
5:19
of
5:20
growth,
5:20
I
5:20
think
5:21
just
5:21
as
5:21
many
5:21
rounds
5:22
of
5:22
funding.
5:22
And
5:22
look,
5:23
when
5:23
people
5:23
ask
5:23
me
5:24
this
5:24
question
5:24
in
5:24
various
5:25
forms,
5:25
I
5:25
always
5:26
answer
5:26
it
5:26
in
5:26
this
5:26
way.
5:27
I
5:27
say,
5:28
for
5:28
me,
5:29
the
5:29
experience
5:29
at
5:30
Climax
5:30
was
5:30
very,
5:31
very
5:31
formative.
5:32
And
5:32
it
5:32
was
5:32
formative
5:33
because
5:34
I
5:34
got
5:34
to
5:35
go
5:35
from
5:35
that
5:36
zero
5:36
to
5:37
one
5:37
experience.
5:38
I,
5:39
as
5:39
somebody
5:40
who
5:40
was
5:40
early,
5:41
I
5:41
got
5:41
to
5:41
grow
5:41
into
5:42
my
5:42
role,
5:42
I
5:42
got
5:43
to
5:43
understand
5:44
what
5:44
it
5:45
was
5:45
all
5:45
about
5:46
and
5:47
I
5:47
really
5:47
got
5:48
to
5:49
take
5:49
on
5:50
senior
5:51
levels
5:51
of
5:52
responsibility
5:54
just
5:54
shy
5:54
of
5:55
being
5:55
responsible
5:55
for
5:56
the
5:56
fate
5:56
of
5:56
the
5:56
entire
5:57
company
5:57
and
5:57
everyone
5:58
who
5:58
worked.
5:58
There
6:00
got
6:01
kind
6:01
of
6:01
all
6:02
the
6:02
exposure
6:03
without
6:03
the
6:04
risk.
6:04
The
6:04
same
6:05
level
6:05
of
6:05
risk
6:05
perhaps.
6:07
So
6:07
you're
6:07
senior
6:08
enough
6:08
to
6:08
make
6:08
big
6:08
decisions,
6:09
but
6:09
not
6:10
senior
6:10
enough
6:10
to
6:10
tank
6:11
the
6:11
company
6:11
is
6:11
the
6:12
very
6:12
blunt
6:12
way
6:12
of
6:12
putting
6:13
it.
6:13
I
6:13
think
6:13
at
6:13
the
6:13
end
6:13
of
6:14
that
6:14
for
6:14
me
6:15
it
6:15
was,
6:15
well,
6:16
now
6:17
I'm
6:17
spending
6:17
X
6:18
amount
6:18
of
6:18
time
6:18
trying
6:19
to
6:19
convince
6:19
y
6:20
person
6:20
of
6:20
Z
6:21
action
6:22
and
6:23
it's
6:23
working,
6:23
but
6:24
not
6:24
as
6:24
I
6:24
want
6:24
or
6:25
it's
6:25
not
6:25
working.
6:26
So
6:26
it's
6:26
time
6:26
to
6:26
take
6:27
the
6:27
risk
6:27
and
6:27
succeed
6:28
or
6:28
fail.
6:28
I
6:29
can
6:29
call
6:29
it
6:29
my
6:29
own.
6:30
And
6:30
so
6:30
that's
6:30
pretty
6:31
much
6:31
how
6:31
it
6:31
went.
6:31
If
6:32
I
6:32
may
6:32
ask,
6:33
given
6:33
that
6:34
you
6:34
launched
6:35
your
6:35
first
6:36
startup
6:37
in
6:37
the
6:37
uk,
6:38
I'm
6:38
guessing,
6:38
right?
6:39
Yeah,
6:39
that's
6:40
right,
6:40
yeah.
6:41
In
6:41
London.
6:41
And
6:42
presumably
6:42
you're
6:43
in
6:43
London
6:43
now
6:43
with
6:44
your.
6:44
Yeah,
6:45
okay,
6:46
we
6:46
are,
6:46
we
6:46
are.
6:47
How
6:47
do
6:47
you
6:48
find?
6:49
Because
6:49
I
6:50
speak
6:50
to
6:50
a
6:50
lot
6:50
of
6:51
founders
6:52
who,
6:52
some
6:52
are
6:53
American,
6:54
some
6:54
are
6:55
European
6:55
based,
6:56
some
6:57
are
6:57
UK
6:57
based
6:58
and
6:59
I'm
6:59
always
7:00
intrigued
7:00
on
7:01
different
7:01
people's
7:02
views
7:02
of
7:02
how.
7:04
I
7:04
suppose
7:05
if
7:06
we.
7:06
Because
7:06
I
7:06
know
7:07
you're
7:07
using
7:08
your
7:08
own
7:08
words,
7:08
you
7:09
want
7:09
to
7:09
create
7:09
a
7:09
gold
7:09
standard
7:10
for
7:11
AI
7:11
learning.
7:12
I
7:12
think
7:12
we
7:12
can
7:12
probably
7:13
agree
7:13
that
7:14
the
7:14
best
7:14
place
7:15
to
7:15
be
7:16
for
7:16
tech
7:17
is
7:17
Silicon
7:18
Valley
7:18
or
7:19
certainly
7:19
America.
7:20
Anyway,
7:21
they're
7:21
very
7:21
attuned
7:22
to
7:22
commercializing
7:23
things.
7:23
And
7:24
what
7:25
do
7:25
you
7:25
think?
7:26
Having
7:26
been
7:27
in
7:27
the
7:27
UK
7:29
and
7:29
built
7:29
in
7:29
the
7:30
uk,
7:30
do
7:30
you
7:30
think
7:31
there
7:31
could
7:31
be
7:31
anything
7:32
about
7:34
the
7:35
startup
7:35
scene
7:36
that
7:36
could
7:36
be
7:36
improved?
7:37
Do
7:37
you
7:38
think
7:38
there's
7:38
anything
7:39
particularly
7:39
great?
7:40
What's
7:40
your
7:40
feeling
7:41
on
7:41
it
7:41
all?
7:43
Yeah,
7:43
it's
7:44
a
7:44
good
7:44
question.
7:44
So
7:45
we
7:45
haven't
7:45
ruled
7:46
out
7:46
a
7:46
Silicon
7:47
Valley
7:47
move.
7:47
Ultimately
7:48
we
7:48
need
7:49
to
7:49
be
7:49
where,
7:49
where
7:49
the
7:50
growth
7:50
is,
7:50
but
7:52
generally
7:52
speaking,
7:54
we
7:54
have
7:55
adopted
7:55
the
7:55
strategy
7:56
of
7:56
moving
7:57
through
7:57
a,
7:58
an
7:58
existing
7:59
distribution
7:59
channels,
8:00
so.
8:00
And
8:00
again,
8:01
we'll
8:01
come
8:01
to
8:01
the
8:01
pivot
8:02
in
8:02
education
8:02
later,
8:03
I'm
8:03
guessing.
8:03
But
8:03
instead
8:04
of
8:04
trying
8:05
to
8:05
build
8:05
Our
8:05
reputation
8:06
from
8:06
scratch
8:07
because
8:07
both
8:07
my
8:07
co
8:08
founder
8:08
Angus
8:08
and
8:09
I,
8:09
we
8:09
come
8:09
from
8:10
commercial
8:10
backgrounds
8:11
and
8:11
we
8:12
particularly,
8:13
I
8:13
can
8:13
point
8:13
to
8:14
lots
8:14
of
8:15
work
8:15
I've
8:15
done
8:16
with
8:16
cutting
8:17
edge
8:17
technology.
8:17
So
8:18
a
8:18
lot
8:18
of
8:18
AI
8:19
and
8:19
groups
8:20
that
8:20
are
8:20
deemed
8:20
particularly
8:21
sensitive.
8:22
So
8:23
you're
8:23
looking
8:23
at
8:24
welfare
8:24
recipients,
8:25
pensioners,
8:26
military
8:27
police,
8:27
that
8:27
sort
8:28
of
8:28
thing.
8:28
So
8:28
lots
8:28
of
8:28
sensitive
8:29
data.
8:29
But
8:30
have
8:30
we
8:30
worked
8:31
in
8:31
education?
8:32
No.
8:32
And
8:32
so
8:33
it
8:33
just
8:33
happened
8:33
to
8:33
be
8:33
that
8:33
the
8:34
distribution
8:34
we
8:35
channel
8:35
that
8:35
we
8:36
found
8:36
best
8:36
worked
8:36
for
8:37
us
8:37
was
8:37
UK
8:38
based
8:39
and
8:39
there
8:39
are
8:40
some
8:40
future
8:41
plans
8:41
effectively
8:42
to
8:42
expand
8:43
into
8:43
the
8:44
us.
8:44
So
8:45
right
8:45
now
8:45
we're
8:45
focusing
8:46
on
8:46
the
8:46
UK
8:47
wider
8:48
continental
8:48
Europe,
8:49
but
8:49
also
8:49
very
8:50
much
8:50
the
8:50
Middle
8:50
East.
8:51
British
8:52
education
8:52
is
8:53
one
8:53
of
8:53
its
8:53
strongest
8:54
exports
8:54
globally
8:55
and
8:56
so
8:56
we
8:56
find
8:56
that
8:56
brand
8:57
works
8:57
really,
8:57
really
8:58
well
8:58
for
8:58
us.
8:58
And
8:59
being
8:59
here,
8:59
being
9:00
based
9:00
here
9:00
in
9:00
London
9:01
or
9:01
in
9:01
the
9:01
UK
9:01
works
9:02
really
9:02
well.
9:03
I
9:03
also
9:03
saw
9:03
an
9:03
Economist
9:04
article
9:04
the
9:04
other
9:05
day
9:05
which
9:05
said
9:06
how
9:06
London
9:07
became
9:07
or
9:08
is
9:08
becoming
9:08
the
9:09
rest
9:09
of
9:09
the
9:09
world's
9:10
startup
9:10
capital.
9:11
So
9:11
outside
9:11
of
9:12
the
9:12
us,
9:12
I
9:13
think
9:13
London
9:13
has
9:14
the
9:14
most
9:14
startup
9:15
funding
9:15
in
9:16
the
9:16
world.
9:16
So
9:16
I
9:16
think
9:16
there's
9:17
three
9:17
or
9:17
four
9:17
cities
9:17
ahead.
9:18
It's
9:18
like
9:18
San
9:19
Francisco,
9:19
New
9:19
York,
9:20
Austin
9:20
and
9:20
maybe
9:21
one
9:21
other
9:21
one
9:21
that
9:21
I
9:22
can't
9:22
remember.
9:22
But
9:23
London
9:23
is
9:23
in
9:23
fourth
9:23
or
9:23
fifth
9:24
place
9:24
globally
9:24
and
9:25
it's
9:25
gaining
9:25
on
9:26
some
9:26
of
9:26
those
9:26
rivals
9:27
in
9:27
the
9:27
States
9:28
as
9:28
well.
9:28
Don't
9:28
want
9:28
to
9:29
go
9:29
too
9:29
much
9:29
into.
9:29
Politics,
9:30
but
9:30
for
9:30
obvious
9:30
reasons
9:31
that's
9:32
not
9:32
this
9:32
podcast.
9:33
No,
9:34
that's
9:34
not
9:34
this
9:34
podcast.
9:35
But
9:35
there
9:36
are
9:36
obvious
9:36
advantages
9:37
to
9:37
being
9:37
outside
9:38
of
9:38
the
9:38
States
9:39
as
9:39
well
9:40
when
9:40
it
9:40
comes
9:40
to
9:41
certain
9:41
deal
9:41
making,
9:42
particularly
9:43
at
9:43
the
9:43
current
9:43
time
9:43
and
9:44
in
9:44
the
9:44
current
9:44
environment.
9:45
But
9:45
never
9:45
say
9:46
never.
9:46
And
9:46
we're
9:47
eyeing
9:47
up
9:47
a
9:47
US
9:48
expansion.
9:48
So
9:49
it'll
9:49
either
9:49
be
9:49
my
9:49
co
9:50
founder
9:50
and
9:50
I,
9:50
but
9:51
also
9:51
someone
9:52
else
9:52
who's,
9:52
who's
9:52
based
9:53
and
9:53
really
9:53
understands
9:53
the
9:54
sector
9:54
who
9:55
might
9:55
help
9:55
expand.
9:56
So
9:56
we
9:56
are
9:57
connected
9:57
to
9:58
quite
9:58
a
9:58
few
9:58
potential
9:59
American
10:00
angel
10:00
investors,
10:01
some
10:01
lovely
10:01
people
10:02
in
10:02
the
10:02
States
10:02
who've
10:03
given
10:03
us
10:03
their
10:03
time
10:03
and
10:04
energy
10:04
and
10:04
so
10:05
we're
10:05
not
10:05
ruling
10:05
anything
10:05
out.
10:06
If
10:06
I
10:06
may
10:06
ask,
10:07
in
10:07
terms
10:08
of
10:08
being
10:08
London
10:09
based,
10:10
is
10:10
that
10:10
more
10:11
to
10:11
do.
10:11
Just
10:12
touching
10:12
on
10:12
with,
10:13
with
10:13
what
10:13
you
10:13
said.
10:13
Is
10:14
that
10:14
more
10:14
to
10:14
do
10:14
with
10:14
the
10:15
fact
10:15
that
10:15
this
10:15
sort
10:16
of
10:16
network
10:17
you're
10:17
breaking
10:17
into
10:18
is
10:18
very
10:18
much
10:19
here
10:19
or.
10:20
Okay,
10:21
yes
10:21
it
10:22
is.
10:22
He's
10:22
nodding
10:23
just
10:23
for
10:23
the
10:23
listeners.
10:24
So
10:25
it's
10:25
a
10:25
bit
10:25
of
10:25
both.
10:25
I
10:26
think
10:26
that
10:26
we,
10:26
my
10:27
co
10:27
founder
10:27
and
10:27
I
10:27
are
10:28
British
10:28
and
10:29
so
10:29
it
10:29
was
10:29
easier
10:29
to
10:30
and
10:30
plan
10:31
a
10:31
move
10:31
across
10:31
the
10:31
pond.
10:32
As
10:32
I
10:32
said,
10:32
the
10:32
British
10:33
brand
10:33
in
10:33
education
10:34
is
10:34
very
10:34
strong
10:35
and
10:35
it's
10:35
a
10:35
logical
10:36
place
10:36
to
10:36
start,
10:36
especially
10:37
with
10:37
our
10:37
distribution
10:38
channel,
10:38
I
10:38
would
10:38
say
10:39
it's
10:39
a
10:39
very
10:39
good
10:39
place
10:40
to
10:40
start
10:41
your
10:41
business
10:42
and
10:42
to
10:42
scale
10:42
it
10:43
to
10:43
a
10:43
particular
10:44
place.
10:44
But
10:44
we
10:44
lack,
10:45
in
10:45
Europe
10:46
generally,
10:46
we
10:47
lack
10:47
the
10:48
sort
10:48
of
10:48
deep
10:49
capital
10:49
that
10:50
you'd
10:50
see
10:50
across
10:51
the
10:51
States
10:51
and
10:52
that
10:52
sort
10:52
of
10:52
sophistication
10:53
of
10:54
investment
10:55
and
10:55
investors
10:57
and
10:57
particularly
10:57
risk
10:58
appetite.
10:58
So
10:59
this
10:59
won't
10:59
be
10:59
anything
11:00
new,
11:00
I
11:00
think
11:00
to
11:00
you
11:01
or
11:01
the
11:01
people
11:02
listening,.
11:02
But
11:02
I
11:03
am
11:03
nodding
11:03
my
11:03
head.
11:05
It's,
11:06
it's
11:06
true,
11:06
you,
11:07
you
11:07
feel
11:07
it.
11:07
But
11:07
we,
11:08
we,
11:08
we,
11:08
we're
11:08
very
11:09
fortunate
11:09
to
11:09
have
11:09
found
11:10
investors
11:10
globally
11:11
and
11:11
people
11:12
who
11:12
are
11:12
very,
11:12
already
11:13
registering
11:13
their,
11:13
their
11:14
interest,
11:14
quite
11:14
a
11:15
few
11:15
in,
11:15
in
11:15
the
11:15
States
11:16
I
11:16
should
11:16
add,
11:16
for,
11:17
for
11:17
our
11:17
next
11:17
round
11:17
of
11:18
funding,
11:18
wherever
11:18
that
11:19
might
11:19
be.
11:19
At
11:19
the
11:19
moment
11:20
you're
11:21
kind
11:21
of
11:21
at
11:21
the
11:22
stage
11:22
where
11:23
you're
11:23
starting
11:23
to
11:25
really,
11:26
I
11:26
guess,
11:27
test,
11:28
test
11:28
the
11:28
product,
11:29
build
11:29
the
11:30
develop
11:31
it
11:32
and
11:32
then
11:33
do
11:33
you
11:33
feel
11:34
you're
11:34
at
11:34
a
11:35
strong
11:35
product
11:35
market
11:36
fit?
11:36
Where,
11:36
where
11:37
do
11:37
you
11:37
feel
11:37
you're
11:37
at
11:37
in,
11:38
in
11:38
the
11:38
process?
11:40
Yeah,
11:40
that's
11:41
pretty
11:42
good
11:42
question.
11:43
So
11:43
we
11:43
initially
11:44
we
11:44
were
11:44
focused
11:45
on
11:45
enterprises,
11:46
so
11:46
mid
11:46
market
11:46
enterprise.
11:47
For
11:47
people
11:48
asking
11:48
what's
11:48
mid
11:48
market?
11:49
You
11:49
can
11:49
define
11:49
it
11:50
in
11:50
many
11:50
ways.
11:50
We
11:50
chose
11:51
to
11:51
define
11:51
it
11:51
as
11:52
so
11:52
companies
11:53
of
11:53
size,
11:54
employee
11:55
count
11:55
49
11:56
to
11:56
500.
11:58
So
11:58
I
11:58
would
11:58
say
11:58
to
11:58
you
11:58
that
11:58
the
11:59
other
11:59
industry
12:00
we're
12:00
targeting
12:01
or
12:01
the
12:01
other
12:01
sort
12:02
of,
12:02
I
12:02
suppose
12:02
area
12:03
or
12:03
domain
12:03
are
12:04
traditional
12:04
industries.
12:05
So
12:06
those
12:06
industries
12:06
that
12:07
have
12:07
had
12:07
capital,
12:08
that
12:08
do
12:08
have
12:08
revenue
12:09
making,
12:09
profit
12:10
making
12:10
businesses
12:11
and
12:11
business
12:12
models
12:12
and
12:13
are
12:13
now
12:13
at
12:13
the
12:14
next
12:14
stage
12:14
of
12:14
wondering
12:15
how
12:15
do
12:15
we
12:15
increase
12:16
productivity?
12:17
The
12:17
taxes
12:18
for
12:18
employers,
12:19
so
12:19
what
12:19
we
12:19
call
12:19
national
12:20
insurance
12:20
in
12:20
the
12:21
UK
12:21
have
12:21
gone
12:22
up.
12:23
So
12:23
how
12:23
do
12:23
we
12:24
address
12:24
that,
12:25
increase
12:25
productivity
12:26
and
12:26
get
12:27
our
12:27
existing
12:27
workers
12:28
their
12:28
maintenance,
12:29
particularly
12:30
a
12:30
low
12:30
hire,
12:30
low
12:30
fire
12:31
environment.
12:31
How
12:32
do
12:32
we
12:33
increase
12:33
the
12:33
productivity
12:34
if
12:34
people
12:34
are
12:34
already
12:35
there?
12:35
So
12:35
the
12:35
people,
12:36
the
12:36
platforms,
12:36
the
12:37
operations,
12:38
what
12:38
we
12:39
found
12:39
is
12:39
that
12:40
as
12:40
we
12:40
spoke
12:41
to
12:41
a
12:41
lot
12:41
of
12:41
those
12:42
people
12:42
and
12:42
businesses,
12:43
they
12:43
were
12:44
very
12:44
keen
12:44
to
12:45
have
12:45
us
12:45
in,
12:45
but
12:46
each
12:46
of
12:46
them
12:47
wanted
12:47
a
12:47
slightly
12:49
different
12:50
thing.
12:50
And
12:51
so
12:51
it's
12:52
been,
12:52
I
12:53
think,
12:53
tricky
12:54
for
12:54
us
12:55
to
12:55
navigate
12:56
enterprise
12:57
and
12:57
our
12:57
pivot
12:58
to
12:58
education
12:59
sort
12:59
of
12:59
came
12:59
with
12:59
this
13:00
idea
13:00
of
13:00
how
13:00
do
13:00
we
13:00
maintain,
13:01
maintain
13:01
B2B
13:02
and
13:02
how
13:02
do
13:03
we
13:03
turn
13:03
the
13:04
advantage
13:04
of
13:04
software,
13:05
which
13:05
is
13:05
predominantly
13:06
a
13:06
scalable
13:07
product,
13:07
at
13:08
the
13:08
same
13:08
standard.
13:09
So
13:09
how
13:10
do
13:10
we
13:10
actually
13:11
turn
13:11
that
13:12
into
13:12
a
13:13
business
13:13
advantage?
13:14
And
13:14
so
13:15
we
13:15
landed
13:15
on
13:16
education.
13:18
And
13:19
that's
13:19
really
13:19
interesting.
13:20
The
13:20
reason
13:20
I
13:20
asked
13:20
that
13:21
question,
13:21
not
13:21
to
13:22
put
13:22
you
13:22
on
13:22
the
13:22
spot
13:22
there
13:22
buddy,
13:23
but
13:23
it's
13:24
more
13:24
to
13:24
do
13:24
with
13:24
the
13:24
fact
13:25
that
13:25
a
13:25
lot
13:25
of
13:26
founders,
13:26
when
13:27
they're
13:27
going
13:27
through
13:28
the
13:28
early
13:28
stage
13:29
of
13:30
developing
13:30
the
13:30
idea,
13:30
testing
13:31
the
13:31
idea,
13:31
having
13:32
those
13:32
conversations,
13:32
trying
13:33
to
13:33
find
13:33
that
13:33
product
13:34
market
13:34
fit.
13:34
There
13:35
is
13:35
a
13:35
tendency,
13:36
and
13:36
you
13:36
touched
13:37
on
13:37
a
13:37
really
13:37
good
13:37
point.
13:37
It
13:38
sounds
13:38
like
13:38
what
13:38
you
13:39
said,
13:39
if
13:39
you
13:39
were
13:39
to
13:39
give
13:39
advice
13:40
here
13:40
would
13:40
be
13:41
for
13:41
people
13:42
who
13:42
are
13:42
maybe
13:42
earlier
13:43
on
13:43
in
13:43
the
13:43
journey
13:44
is
13:45
if
13:45
you're
13:46
trying
13:46
to
13:46
build
13:46
something
13:46
scalable,
13:48
go
13:48
figure.
13:49
Trying
13:49
to
13:50
work
13:50
with
13:50
industries
13:51
where
13:51
you
13:51
soon
13:52
realize
13:52
that
13:52
you
13:52
need
13:53
to
13:54
tailor
13:54
the
13:55
product
13:56
to
13:56
every
13:56
individual
13:57
company,
13:59
you
13:59
will
13:59
soon
14:00
have
14:00
a
14:00
scale
14:01
problem,
14:02
a
14:02
massive
14:02
scale
14:02
problem.
14:03
I
14:03
think
14:03
that's
14:04
what
14:04
I'd
14:04
call
14:04
the
14:04
Palantir
14:05
problem.
14:05
So
14:06
Palantir
14:06
is
14:06
very
14:07
successful
14:07
today.
14:08
But
14:08
if
14:08
people
14:08
aren't
14:09
familiar
14:10
with
14:10
Palantir
14:11
and
14:11
you
14:11
know,
14:11
their
14:11
stock
14:11
price
14:12
has
14:12
really
14:12
only
14:12
shot
14:12
up
14:13
in
14:13
the
14:13
last
14:13
sort
14:13
of
14:13
couple
14:14
of
14:14
years
14:14
significantly.
14:15
But
14:15
if
14:15
people
14:15
are
14:16
not
14:16
familiar
14:16
with
14:16
Palantir,
14:17
they
14:17
have
14:17
been
14:17
around
14:18
in
14:18
some
14:19
form
14:19
or
14:19
another
14:20
since
14:20
I
14:20
think
14:21
it's
14:21
2004
14:22
or
14:23
so.
14:23
But
14:23
worth
14:24
checking
14:24
that
14:24
they
14:25
started
14:25
with
14:25
being
14:25
sort
14:26
of
14:26
business
14:26
analytics
14:27
and
14:27
then
14:28
moved
14:28
into
14:29
defense
14:30
and
14:30
similar
14:31
industry.
14:32
So
14:32
national
14:32
security,
14:33
however
14:34
that's
14:34
defined
14:35
and
14:35
effectively
14:36
only
14:36
recently
14:37
has
14:37
Palantir
14:38
said
14:38
with
14:38
pride
14:39
that
14:39
they
14:39
now
14:40
standardize
14:40
their
14:41
updates
14:41
across
14:42
some
14:42
aspects
14:43
of
14:43
their
14:43
code
14:43
and
14:43
their
14:44
products.
14:45
And
14:45
if
14:45
you
14:45
come
14:45
from
14:46
that,
14:46
from
14:46
a
14:46
traditional
14:46
software
14:47
background
14:47
or
14:47
a
14:48
SaaS
14:48
background,
14:48
you
14:49
might
14:49
think
14:49
that
14:49
is
14:50
crazy,
14:50
that
14:51
they're
14:51
proud
14:51
of
14:51
that,
14:52
but
14:52
they're
14:52
proud
14:52
of
14:52
it
14:53
because
14:53
effectively
14:54
every
14:54
one
14:55
of
14:55
their
14:55
deployments
14:56
into
14:57
a
14:57
domain
14:57
of
14:57
national
14:58
security
14:58
or
14:58
whatever
14:59
is
14:59
its
14:59
own
14:59
product.
15:00
Palantir
15:01
has
15:01
that
15:01
footprint.
15:02
It's
15:02
just
15:02
found
15:02
it
15:03
very
15:03
difficult
15:03
to
15:03
scale.
15:04
It's
15:04
got
15:04
stuck
15:05
into
15:05
those
15:05
niche
15:06
areas.
15:06
It's
15:06
made
15:07
itself
15:07
very
15:07
sticky
15:08
to
15:08
use
15:08
traditional
15:09
terms
15:10
here
15:10
for
15:11
software
15:11
or
15:12
business
15:12
models.
15:13
But
15:13
ultimately
15:14
it's
15:14
hard
15:15
to
15:15
scale
15:15
that
15:16
business
15:16
because
15:16
you
15:16
can't
15:17
go
15:17
and
15:17
take
15:17
the
15:17
lessons
15:18
learned
15:18
necessarily
15:19
from
15:19
one
15:19
very
15:19
secure
15:20
domain
15:20
immediately
15:21
into
15:21
another.
15:21
And
15:22
so
15:22
in
15:22
education
15:23
and
15:23
I'll
15:23
stop
15:23
in
15:23
in
15:24
a
15:24
second
15:24
so
15:24
you
15:24
can,
15:25
you
15:25
can
15:25
ask
15:26
what
15:26
you're
15:27
saying
15:27
is.
15:27
Really
15:27
interesting
15:28
to
15:28
the.
15:28
It
15:29
will
15:29
be
15:29
to
15:29
the
15:29
audience.
15:30
So
15:30
perfect,
15:31
carry
15:31
on.
15:32
So
15:33
I,
15:34
I
15:34
think
15:34
for
15:34
us
15:35
to
15:35
come
15:35
back
15:35
to
15:35
what
15:35
I
15:35
said
15:35
previously,
15:36
then
15:36
how
15:36
do
15:37
you
15:37
turn
15:37
that
15:37
scalability
15:38
problem
15:39
into
15:39
something
15:39
that
15:39
works
15:40
for
15:40
you?
15:40
We
15:40
came
15:41
across
15:41
a,
15:42
there
15:42
was
15:42
a
15:42
sort
15:43
of
15:44
investment
15:44
arm
15:45
or
15:45
a
15:45
small
15:45
sort
15:46
of
15:46
boutique
15:47
investment
15:47
company
15:48
that
15:48
connected
15:49
us
15:49
as
15:50
a
15:50
company,
15:50
my
15:50
co
15:50
founder
15:51
and
15:51
I,
15:51
to
15:52
this
15:52
distribution
15:52
channel
15:53
that
15:53
we've
15:53
got
15:53
this
15:54
agreement
15:54
with
15:55
and
15:55
they
15:56
have
15:56
spent
15:56
the
15:57
better
15:57
part
15:57
of
15:58
20
15:58
something
15:59
years
15:59
effectively
16:00
building
16:00
up
16:00
a
16:00
reputation
16:01
with
16:02
schools
16:02
across,
16:03
across
16:03
the
16:03
country
16:04
and
16:04
what
16:05
schools,
16:05
and
16:06
I
16:06
should
16:06
say
16:07
now
16:08
these
16:08
terms
16:09
mean
16:09
different
16:09
things
16:10
across
16:10
the
16:10
pond.
16:10
So
16:11
I'll
16:11
be
16:11
very
16:11
specific.
16:12
Public
16:12
and
16:12
private.
16:14
Public
16:14
and
16:14
private.
16:14
That's
16:15
it.
16:15
Private
16:16
or
16:16
independent
16:17
schools.
16:17
So
16:17
that
16:17
is
16:18
fee
16:18
paying
16:18
schools
16:19
where
16:19
the
16:19
parents
16:20
pick
16:20
up.
16:21
Not
16:21
paid
16:21
for
16:21
by
16:21
the
16:22
state.
16:23
Not
16:23
paid
16:23
for
16:23
by
16:23
the
16:24
state.
16:24
Exactly.
16:24
Not
16:25
publicly
16:25
funded.
16:26
So
16:27
first
16:27
of
16:27
all
16:27
you
16:27
think
16:28
education
16:28
and
16:29
you
16:29
think,
16:29
all
16:29
right,
16:29
we're
16:30
going
16:30
to
16:30
scale
16:30
something
16:31
across
16:31
the
16:31
sector
16:32
and
16:32
then
16:32
you
16:32
think,
16:33
well
16:33
who's
16:33
going
16:33
to
16:33
pay
16:34
for
16:34
it?
16:34
Well,
16:35
you
16:35
need
16:35
the
16:36
schools
16:36
with
16:36
the
16:37
most
16:37
money
16:37
to
16:37
pay
16:37
for
16:37
it.
16:38
Are
16:38
they
16:38
willing
16:38
to
16:38
pay
16:39
for
16:39
it?
16:39
Well,
16:40
put
16:40
yourselves,
16:41
put
16:41
yourself
16:41
in
16:42
the
16:42
shoes
16:42
of
16:42
a
16:42
parent
16:43
who's
16:43
sending
16:44
1,
16:44
2,
16:44
3,
16:45
however
16:45
many
16:45
children
16:45
it
16:46
might
16:46
be
16:46
to
16:46
these
16:46
privately
16:47
funded
16:47
schools.
16:48
Sometimes
16:48
these
16:49
schools,
16:49
particularly
16:50
in
16:50
Britain,
16:51
their
16:51
fees
16:51
go
16:52
up
16:52
to,
16:52
you
16:52
know,
16:53
depending
16:53
on
16:53
which
16:53
school
16:53
it
16:54
is,
16:54
about
16:55
60,
16:55
70,
16:56
80,000
16:57
pounds
16:57
US
16:57
dollars
16:59
a
16:59
year
17:00
per
17:00
child.
17:02
So
17:02
you
17:03
know,
17:03
you
17:03
are
17:03
likely
17:03
to
17:04
have
17:04
the
17:04
children
17:04
of
17:05
everyone
17:05
from
17:06
sort
17:06
of
17:06
diplomats,
17:07
high
17:07
ranking
17:08
officials
17:08
in
17:08
certain
17:09
countries,
17:09
big
17:10
bankers
17:11
in,
17:11
in
17:11
one
17:11
sort
17:12
of
17:12
learning
17:12
environment,
17:13
you
17:13
know,
17:13
where,
17:13
where
17:13
they
17:14
are.
17:14
Sure.
17:14
Or
17:15
they
17:15
expect
17:15
their
17:15
children
17:16
to
17:16
be
17:16
taught
17:17
at
17:17
an
17:17
elite
17:17
level
17:17
level.
17:18
So
17:18
what
17:18
is
17:19
an
17:19
elite
17:19
level
17:19
today?
17:20
Well,
17:21
ultimately,
17:22
what
17:22
do
17:22
schools
17:23
exist
17:23
to
17:23
do?
17:23
Are
17:24
they
17:24
there
17:24
to
17:24
educate
17:25
or
17:25
are
17:25
they
17:25
there
17:25
to
17:26
educate
17:26
and
17:26
then
17:26
teach
17:27
about
17:27
the
17:27
future
17:27
of
17:28
work?
17:28
And
17:29
that's
17:29
the
17:29
big
17:29
bet
17:29
we've
17:30
made
17:30
that
17:30
they're
17:31
there
17:31
to
17:31
do
17:31
that
17:31
second
17:31
part
17:32
as
17:32
well.
17:33
And
17:35
sadly
17:36
we're
17:36
at
17:36
a
17:36
point
17:36
where
17:37
I'm
17:37
not
17:37
going
17:37
to
17:38
besmirch
17:38
the
17:39
name
17:39
of
17:39
one
17:39
of
17:39
the
17:39
big
17:40
frontier
17:40
models
17:41
companies,
17:41
but
17:42
I
17:42
will
17:42
say
17:42
that
17:43
at
17:43
the
17:43
beginning
17:43
of
17:43
this
17:44
journey
17:44
they've
17:45
sort
17:45
of
17:45
said
17:45
our
17:46
product
17:46
is
17:46
safe.
17:47
Here
17:47
are
17:48
controls,
17:49
parental
17:49
controls.
17:50
So
17:50
you
17:51
parents,
17:51
you
17:51
can
17:51
control
17:52
how
17:52
your
17:52
children
17:52
is
17:53
using
17:53
this,
17:54
your
17:54
child,
17:54
sorry,
17:55
is
17:55
using
17:55
this
17:55
product.
17:56
They,
17:56
those
17:57
parental
17:57
controls
17:57
have
17:57
then
17:58
been
17:58
broken
17:58
inside
17:58
of
17:59
one
17:59
day,
17:59
24
18:00
hours
18:00
later
18:01
they're
18:01
in
18:01
pieces.
18:02
You
18:02
then
18:03
have
18:03
that
18:04
same
18:05
company
18:06
saying,
18:06
here
18:06
are
18:06
some
18:07
resources
18:07
there
18:07
for
18:08
educators,
18:08
right?
18:08
Here's
18:09
how
18:09
to
18:09
get
18:09
it
18:09
into,
18:10
get
18:10
the
18:11
proper,
18:11
responsible
18:12
AI
18:12
use,
18:13
ethical
18:13
AI
18:13
use
18:14
into
18:14
the
18:14
minds
18:14
of
18:14
children.
18:15
They've
18:16
then
18:16
said
18:16
as
18:17
a
18:17
company,
18:17
separate
18:18
product
18:18
offering,
18:18
but
18:18
same
18:19
company,
18:19
you
18:19
know,
18:20
some
18:20
number
18:21
of
18:21
weeks
18:21
later,
18:22
you
18:22
know
18:22
what,
18:23
we
18:24
think
18:24
that
18:24
adults,
18:25
we
18:25
should
18:25
treat
18:25
people
18:26
like
18:26
adults
18:26
when
18:27
they
18:27
use
18:27
this
18:27
software.
18:28
And
18:28
therefore
18:29
we're
18:29
going
18:30
to
18:30
allow
18:30
people
18:30
to
18:31
use
18:31
our
18:32
AI
18:32
to
18:32
generate
18:33
adult
18:33
content.
18:34
Now
18:34
you
18:35
take
18:35
these
18:35
things
18:35
together,
18:36
parental
18:36
controls
18:37
that
18:37
don't
18:37
work
18:38
or
18:38
that
18:38
are
18:38
broken,
18:39
you've
18:39
got
18:39
resources
18:40
for
18:40
teachers
18:41
being
18:41
provided
18:42
from
18:42
a
18:42
place
18:42
of
18:43
not
18:43
really
18:43
that
18:44
much
18:44
experience.
18:45
This
18:45
is
18:45
not
18:45
an
18:45
education
18:46
company
18:46
or
18:47
one
18:47
really
18:47
that's
18:48
shown
18:48
much
18:48
interest
18:49
in
18:49
schooling
18:50
responsibly
18:51
and
18:51
the
18:51
generation
18:52
of
18:52
adult
18:52
content
18:53
or
18:53
from
18:53
the
18:53
same
18:53
platform
18:54
or
18:54
same
18:55
company
18:56
doesn't
18:56
really
18:56
scream
18:57
safeguarding
18:58
in
18:58
schools.
18:59
It
18:59
doesn't
18:59
really
18:59
scream
19:00
safeguarding
19:00
in
19:01
children.
19:01
And
19:02
so
19:02
some
19:02
schools
19:02
have
19:03
seen
19:03
this
19:03
problem
19:03
pretty
19:04
swiftly
19:04
and
19:04
have
19:05
thought,
19:05
now,
19:06
do
19:06
we
19:06
engage
19:06
with
19:07
this
19:07
product
19:07
or
19:08
do
19:08
we
19:08
say
19:09
we're
19:09
already
19:09
pretty
19:10
busy
19:10
with
19:10
what
19:10
we've
19:11
got
19:11
to
19:11
deliver
19:11
as
19:12
part
19:12
of
19:12
our
19:12
national
19:13
curriculum,
19:13
as
19:14
part
19:14
of
19:14
our
19:15
agreement
19:15
with
19:15
the
19:15
government,
19:16
the
19:16
national
19:16
government,
19:17
with
19:17
local
19:17
councils,
19:18
what
19:18
have
19:19
you
19:19
with
19:19
national
19:19
priorities?
19:20
And
19:21
do
19:21
you
19:22
know
19:22
what?
19:22
I
19:22
don't
19:22
know
19:23
how
19:23
to
19:23
teach
19:23
this
19:23
subject.
19:24
Hands
19:24
off.
19:25
We're
19:25
just
19:25
not
19:25
going
19:25
to
19:26
teach
19:26
it
19:26
at
19:26
school.
19:27
Now,
19:29
those
19:29
of
19:29
us
19:29
who
19:30
aren't
19:30
intimately
19:30
involved
19:31
in
19:31
education,
19:31
we
19:32
can't
19:32
say
19:32
that
19:32
that's
19:33
lazy
19:33
or
19:33
whatever,
19:34
because
19:34
there
19:34
are
19:34
lots
19:35
of
19:35
competing
19:35
priorities.
19:36
What
19:36
we
19:37
can
19:37
say
19:37
fairly
19:38
is
19:38
that
19:38
we
19:38
think
19:38
it's
19:39
an
19:39
abdication
19:39
of
19:39
responsibility
19:41
at
19:41
schools
19:42
not
19:42
to
19:42
teach
19:42
about
19:43
the.
19:44
Whatever
19:44
it's
19:44
now
19:44
been
19:45
called
19:45
the
19:45
Third
19:45
Industrial
19:46
Revolution,
19:47
right?
19:47
If
19:47
maybe
19:48
electricity
19:49
or
19:49
the
19:49
fourth
19:49
Industrial
19:50
Revolution,
19:50
if
19:51
you
19:51
count
19:51
the
19:51
Industrial
19:52
Revolution,
19:53
electricity,
19:53
the
19:54
Internet
19:54
and
19:55
then
19:55
AI.
19:55
And
19:56
so
19:56
we've
19:56
gotten
19:56
to
19:56
a
19:57
point
19:57
really,
19:57
where
19:57
if
19:57
you're
19:58
a
19:58
parent
19:58
now,
19:58
coming
19:59
back
19:59
to
19:59
this
19:59
idea,
20:00
looking
20:00
at
20:00
the
20:00
future
20:00
of
20:01
work,
20:01
is
20:01
your
20:02
school,
20:02
the
20:02
school
20:02
that
20:02
you're
20:03
paying
20:03
up
20:03
to
20:03
120,
20:04
130,000
20:05
USD
20:06
for
20:06
a
20:07
year
20:08
is
20:08
that.
20:09
Is
20:09
your
20:10
school
20:11
teaching
20:11
your
20:11
child
20:12
how
20:13
to
20:13
use
20:13
the
20:13
latest
20:14
and
20:14
greatest
20:14
technology
20:15
responsibly
20:16
and
20:17
effectively,
20:18
so
20:18
that
20:18
tomorrow,
20:19
when
20:19
they
20:19
go
20:19
out
20:20
into
20:20
the
20:20
job
20:20
market,
20:21
they
20:22
look
20:22
to
20:23
be
20:23
a
20:23
stronger
20:23
candidate,
20:24
they
20:25
look
20:25
to
20:25
be
20:26
engaged
20:26
with
20:27
the
20:27
technology
20:27
of
20:28
what
20:28
will
20:28
then
20:28
be
20:28
today
20:29
rather
20:29
than
20:30
the
20:30
technology
20:30
of
20:31
tomorrow?
20:32
I
20:32
think
20:33
what
20:33
you're
20:33
touching
20:33
on
20:34
is
20:34
fantastic
20:34
because
20:35
I
20:35
read
20:35
an
20:36
article
20:36
the
20:36
other
20:36
day
20:37
after
20:37
our
20:37
first
20:37
conversation,
20:38
randomly,
20:39
I
20:39
swear,
20:41
you
20:41
know,
20:41
you
20:42
know
20:42
when
20:42
stuff
20:42
pops
20:42
up
20:42
and
20:43
you're
20:43
like,
20:43
I
20:43
swear
20:43
I'm
20:43
being
20:44
listened
20:44
to.
20:44
Yes,
20:46
I
20:46
may
20:46
or
20:46
may
20:46
not
20:47
be,
20:47
I
20:47
can't
20:48
say
20:48
for
20:48
sure.
20:49
But
20:49
the
20:50
interesting
20:50
point
20:51
was,
20:51
I
20:51
believe
20:52
in
20:52
China,
20:53
they're
20:53
now
20:53
instilling
20:54
AI
20:55
education
20:55
from
20:55
the
20:56
age
20:56
of
20:56
five.
20:58
And
20:59
of
20:59
course
20:59
the
21:00
uk
21:00
just
21:00
to
21:01
take
21:01
one
21:01
country,
21:02
they
21:02
are
21:03
obviously
21:03
facing
21:04
a
21:04
huge
21:05
deficit.
21:05
Same
21:05
in
21:05
the
21:06
us,
21:06
A
21:07
slowing
21:07
economy,
21:08
years
21:08
of
21:08
high
21:09
inflation,
21:10
they
21:10
need
21:11
more
21:11
tax
21:11
revenues,
21:12
spending
21:12
is
21:13
out
21:13
of
21:13
control.
21:14
So
21:15
fundamentally,
21:16
it's
21:17
really
21:17
hard
21:18
for.
21:20
I
21:20
feel
21:20
like
21:21
it's
21:21
your
21:21
timing
21:22
to
21:22
really
21:22
make
21:23
a
21:23
positive
21:23
impact
21:24
is
21:24
right,
21:25
because,
21:25
you
21:25
know,
21:26
were
21:26
not
21:26
just,
21:27
you
21:27
know,
21:27
in
21:27
traditional
21:28
sense
21:29
30
21:29
years
21:29
ago,
21:29
40
21:30
years
21:30
ago,
21:30
when
21:31
your
21:31
child
21:31
came
21:32
out
21:32
of
21:32
high
21:33
school,
21:33
college,
21:34
whatever
21:34
university
21:35
they
21:35
would
21:35
then
21:36
Be
21:36
against
21:36
whoever
21:37
applies
21:38
maybe
21:38
locally
21:39
or
21:39
in
21:40
the
21:40
same
21:40
county
21:41
or
21:41
whatever,
21:42
against
21:42
other
21:42
people
21:43
with
21:43
that
21:43
set
21:43
of
21:43
skills.
21:44
Right.
21:44
So
21:45
it
21:45
tended
21:45
to
21:45
be
21:45
quite
21:46
regional,
21:47
quite
21:47
geographic.
21:48
We're
21:48
now
21:48
in
21:48
a
21:48
world
21:49
where
21:49
it
21:50
seems
21:50
like
21:50
AI,
21:51
Tell
21:51
me
21:51
if
21:51
I'm
21:51
wrong,
21:52
but
21:52
AI
21:53
kind
21:53
of
21:54
levels
21:54
it
21:54
in
21:54
the
21:55
sense
21:55
that
21:55
I
21:55
know
21:56
you
21:56
speak
21:56
four
21:57
languages
21:57
pretty
21:58
well
21:58
from
21:59
your
21:59
God.
22:00
Yeah,
22:00
it'll
22:00
probably
22:01
be
22:01
three
22:01
now.
22:02
Yeah,
22:03
A
22:03
lack
22:03
of
22:03
time,
22:04
I
22:04
think,
22:04
when
22:04
you,
22:05
when
22:05
you
22:05
make
22:05
it
22:05
to.
22:06
But
22:06
the
22:06
interesting
22:07
point
22:07
is
22:08
like,
22:08
it
22:09
sounds
22:09
like
22:09
what
22:09
you're
22:10
suggesting
22:10
is
22:10
we
22:11
need
22:11
to
22:11
like
22:12
act
22:12
now
22:12
to
22:13
get
22:13
this
22:13
in
22:13
place
22:14
to
22:14
make
22:15
sure
22:15
that
22:15
people
22:15
are
22:15
prepared.
22:16
Because
22:16
it's
22:16
not
22:16
just
22:17
being
22:17
prepared
22:17
to
22:17
be
22:18
a
22:18
good
22:19
engineer
22:20
or
22:20
a
22:20
good
22:20
business
22:21
person
22:21
or
22:21
whatever.
22:22
It's
22:23
about
22:23
being
22:23
good
22:24
with
22:24
the
22:24
tools
22:25
that
22:25
make
22:26
you
22:27
able
22:27
to
22:27
do,
22:28
you
22:29
know,
22:29
three
22:29
times,
22:30
you
22:31
know,
22:31
more
22:31
work
22:31
or
22:32
whatever
22:32
the
22:32
number
22:32
is,
22:33
and
22:33
be
22:33
the
22:34
best
22:34
possible
22:35
hire
22:35
and
22:36
putting
22:36
your
22:36
kids
22:37
in
22:37
the
22:37
best
22:37
possible
22:38
place.
22:39
So
22:40
if
22:40
you're.
22:41
He's
22:41
nodding.
22:42
For
22:42
those
22:43
listening,
22:43
I'm
22:43
nodding.
22:44
So
22:44
the,
22:44
the
22:44
interesting
22:45
question
22:45
is
22:46
how
22:46
did
22:46
you
22:47
realize
22:47
that
22:47
the
22:47
way
22:48
that
22:48
AI
22:48
was
22:49
being
22:49
taught
22:49
is
22:50
fundamentally
22:51
broken
22:51
or
22:52
non
22:52
existent?
22:53
And
22:53
like,
22:53
how
22:54
did
22:54
you.
22:54
I
22:54
believe
22:55
you're
22:55
focusing
22:55
on
22:56
micro
22:56
learning,
22:57
is
22:57
that
22:57
correct?
22:57
Yeah,
22:58
we,
22:59
we
22:59
are,
22:59
yeah.
23:00
So
23:00
we're,
23:00
we're
23:00
focusing
23:01
on
23:01
microlearning.
23:02
I
23:02
think
23:02
there's
23:02
a
23:02
fairly
23:03
short
23:03
answer
23:03
to
23:03
your
23:04
question,
23:04
actually,
23:04
which
23:04
is
23:05
it's
23:05
not
23:05
being
23:05
taught
23:06
and
23:07
where
23:07
it
23:08
is.
23:08
So
23:08
just
23:09
to
23:09
touch
23:09
on
23:09
what
23:09
you
23:09
were
23:10
saying,
23:10
it's.
23:11
It's
23:12
being
23:12
taught
23:13
in
23:14
education
23:14
environments
23:15
with
23:15
more
23:15
resources.
23:16
And
23:17
so
23:19
I
23:19
think
23:19
in
23:19
education
23:20
itself,
23:20
it's
23:21
barely
23:21
being
23:21
taught
23:21
or
23:21
it's
23:22
not
23:22
being
23:22
taught.
23:22
And
23:22
then
23:23
where
23:23
it
23:23
is
23:23
being
23:23
taught,
23:24
it's.
23:24
Yeah.
23:25
In
23:25
places
23:25
with
23:25
more
23:25
resources.
23:26
So
23:27
sort
23:27
of
23:27
far
23:27
from
23:28
being
23:28
this
23:29
great
23:29
leveler,
23:30
there
23:31
are
23:31
genuine
23:32
worries
23:32
that
23:32
AI
23:33
will
23:33
end
23:34
up
23:34
being
23:34
a
23:35
tool
23:36
used
23:36
for
23:37
entrenchment,
23:38
effectively.
23:39
So
23:39
the
23:39
entrenchment
23:40
of
23:40
current
23:41
social
23:42
class
23:42
or
23:43
socioeconomic
23:44
classes.
23:45
And
23:45
so
23:46
I
23:46
think
23:46
this
23:46
goes
23:47
beyond
23:48
private
23:48
schools.
23:49
Right.
23:49
So
23:49
I
23:49
know,
23:49
Mike,
23:50
you
23:50
and
23:50
I
23:50
were
23:50
talking
23:50
about
23:51
this
23:51
a
23:51
few
23:51
weeks
23:51
ago,
23:51
but
23:52
just
23:52
for
23:53
the
23:53
listeners,
23:54
it
23:54
goes
23:54
beyond
23:54
this.
23:55
So
23:55
I
23:55
think
23:55
we
23:55
want
23:56
to
23:56
start
23:56
with
23:56
private
23:56
schools
23:57
and
23:57
then
23:57
we
23:58
effectively
23:58
want
23:59
to
23:59
engage
23:59
with
24:00
the
24:00
various
24:00
departments
24:01
for
24:01
education
24:01
in
24:02
the
24:02
us,
24:02
the
24:03
uk,
24:04
potentially
24:05
continental
24:05
Europe,
24:06
some
24:06
of
24:06
the
24:06
regional
24:07
bodies
24:07
in
24:08
the
24:08
Middle
24:08
east
24:08
as
24:08
well,
24:09
for
24:09
a
24:09
starter.
24:11
And
24:11
say,
24:12
look,
24:13
we've
24:14
implemented
24:15
this
24:15
at
24:15
fee
24:15
paying
24:16
schools,
24:17
we've
24:17
collected
24:18
a
24:18
ton
24:18
of
24:19
aggregate
24:19
analytics,
24:20
all
24:20
anonymized
24:21
and
24:21
so
24:21
on.
24:21
But
24:22
here's
24:23
how
24:23
it's
24:23
helping.
24:24
Here's
24:24
actively
24:25
where
24:25
we
24:25
were
24:25
at
24:26
the
24:26
beginning
24:26
of
24:26
a
24:26
lot
24:26
of
24:26
these
24:27
journeys.
24:27
Here's
24:28
where
24:28
we
24:28
are
24:28
now.
24:28
And
24:29
so
24:29
we've
24:29
gone
24:29
from
24:30
nothing
24:31
slash
24:31
ad
24:32
hoc
24:32
training
24:33
slash
24:33
best
24:34
efforts
24:34
to
24:35
a
24:35
structured
24:36
program
24:37
that
24:37
dovetails
24:38
very
24:39
nicely
24:40
with
24:40
the
24:40
existing
24:41
curriculum.
24:41
And
24:42
so
24:42
it's
24:42
embedded
24:43
at
24:43
every
24:43
stage.
24:45
And
24:45
we
24:45
effectively
24:46
want
24:46
to
24:46
take
24:46
that
24:46
to
24:47
these
24:47
departments
24:47
of
24:47
education
24:48
or
24:48
these
24:48
regional
24:49
bodies
24:49
and
24:50
say
24:51
it's
24:51
not
24:52
fair,
24:52
is
24:52
it?
24:52
Why,
24:53
why
24:53
should
24:53
only
24:54
the
24:54
fee
24:54
paying
24:55
schools
24:55
have
24:55
this?
24:55
The
24:56
reason
24:56
we
24:56
don't
24:56
start
24:56
with
24:57
that
24:57
is,
24:57
is
24:57
obvious.
24:58
Commercially
24:58
it
24:58
doesn't
24:59
make
24:59
sense
25:00
for
25:00
us.
25:01
And
25:01
secondly,
25:02
it
25:02
requires
25:02
a
25:02
very
25:02
different
25:03
business
25:03
model
25:03
and
25:04
it
25:04
requires
25:04
engagement
25:05
at
25:05
the
25:05
national,
25:06
potentially
25:06
supranational
25:07
level
25:08
as
25:08
well
25:08
as
25:09
depending
25:09
on
25:09
which
25:09
blocks.
25:10
Here
25:10
we're
25:10
talking
25:11
about
25:11
blocks
25:11
of
25:12
countries,
25:12
but
25:12
also
25:13
at
25:13
the,
25:13
you
25:13
know,
25:13
zooming
25:14
down
25:14
or
25:15
lower
25:15
level
25:16
at
25:16
the
25:16
council
25:17
level
25:17
and
25:17
whatnot.
25:18
So
25:18
it
25:18
requires
25:19
a
25:19
very
25:19
different
25:20
type
25:20
of
25:21
approach.
25:22
And
25:22
yeah,
25:23
I
25:24
mean,
25:24
yeah,
25:24
government
25:25
tenders
25:25
contracts
25:26
are
25:26
renowned
25:26
to
25:27
be
25:29
like,
25:29
you
25:30
know,
25:30
I
25:30
don't
25:30
know,
25:30
I'm
25:31
trying
25:31
to
25:31
think
25:31
of,
25:32
I
25:32
don't
25:32
know,
25:32
a
25:32
Red
25:32
Bull
25:33
event
25:33
where
25:34
you
25:34
have
25:34
to
25:34
climb
25:35
a
25:35
mountain,
25:35
you
25:35
know,
25:36
I
25:36
don't
25:36
know.
25:37
But
25:37
this
25:38
is
25:38
it.
25:38
Maybe
25:38
you're
25:39
watching
25:39
the,
25:39
what's
25:40
his
25:40
name?
25:40
Honold,
25:41
the
25:42
free
25:42
solo
25:42
guy
25:43
scale.
25:43
The,
25:44
the
25:44
Singapore.
25:45
I
25:46
haven't
25:46
seen
25:46
that
25:46
yet.
25:47
No,
25:47
on
25:47
Netflix,
25:48
right?
25:48
On
25:49
Netflix,
25:49
Yeah.
25:49
He
25:50
did
25:50
it
25:50
live.
25:50
Yeah.
25:51
Slightly,
25:52
slightly
25:52
insane,
25:53
man.
25:53
It's
25:53
fun
25:53
when,
25:54
when
25:54
you
25:54
talk
25:55
to
25:55
other
25:55
founders
25:56
about
25:56
the
25:57
sort
25:57
of
25:57
journey
25:57
of
25:58
insanity
25:58
you've
25:58
chosen
25:59
for
25:59
yourself,
25:59
you
25:59
remember
26:00
that
26:00
you
26:00
humble
26:01
yourself
26:01
quite
26:01
quickly
26:02
by
26:02
remembering
26:02
there
26:03
are
26:03
people
26:03
who
26:03
genuinely
26:04
put
26:04
their
26:04
lives
26:04
at
26:04
danger
26:05
by
26:05
being
26:05
even
26:06
more
26:06
insane.
26:07
But
26:08
just
26:08
to
26:08
come
26:08
back
26:09
to
26:09
your
26:09
point
26:10
quickly,
26:11
this
26:11
procurement
26:12
process
26:12
stuff,
26:12
this
26:13
is
26:13
what
26:13
we
26:13
tackled
26:13
in
26:14
my
26:14
first
26:14
AI
26:15
startup
26:15
in
26:16
we
26:16
are.
26:16
Right.
26:16
So
26:16
you
26:17
would
26:17
have.
26:22
Companies
26:22
spending
26:23
up
26:23
to
26:23
a
26:24
sort
26:24
of
26:25
million
26:25
dollars
26:26
on
26:27
some
26:27
of
26:27
their
26:27
chunkier
26:28
procurement
26:28
proposals
26:29
with
26:29
the
26:29
guarantee
26:30
they
26:30
would
26:30
only
26:30
be
26:31
in
26:31
the
26:31
final
26:31
three.
26:32
And
26:32
we
26:32
used
26:32
to
26:32
focus
26:33
on
26:33
civil
26:33
engineering
26:34
companies.
26:34
So
26:34
you
26:35
know,
26:35
it's
26:35
not
26:35
software
26:36
civil
26:36
engineering.
26:37
You
26:37
can't
26:37
make
26:37
a
26:37
bridge
26:38
that
26:38
collapses
26:38
in
26:39
first
26:39
year
26:39
of
26:39
opening.
26:40
That
26:40
bridge
26:40
has
26:41
got
26:41
to
26:41
stand
26:41
for
26:41
30
26:41
years
26:42
or
26:42
more
26:42
and
26:42
you've
26:43
got.
26:43
30
26:43
day
26:44
money
26:44
back.
26:44
Guarantee
26:45
is
26:45
relevant.
26:47
Well,
26:47
so,
26:48
and,
26:48
and,
26:49
and
26:49
so
26:49
yeah,
26:49
this
26:50
is
26:50
it,
26:50
you
26:50
know,
26:50
and
26:51
those,
26:51
those
26:52
proposals
26:52
used
26:52
to
26:52
be
26:53
up
26:53
to
26:53
3,000
26:53
pages
26:54
long
26:54
with
26:55
crazy
26:55
amounts
26:56
of,
26:56
you
26:56
know,
26:57
detail.
26:57
It
26:58
contained
26:58
in
26:58
some
26:58
of
26:59
these
26:59
pages
27:00
and
27:00
whether
27:00
that
27:00
company
27:01
won
27:01
or
27:01
lost
27:01
that
27:02
contract
27:02
would
27:03
be,
27:03
sometimes
27:04
they
27:04
would
27:04
win
27:04
or
27:04
lose
27:05
it
27:05
in
27:05
the
27:05
margins
27:06
of
27:06
a,
27:06
you
27:07
know,
27:07
zero
27:07
point
27:08
Something
27:08
percent.
27:09
They
27:09
had
27:10
some
27:10
sort
27:10
of
27:10
like
27:11
our
27:11
strange
27:12
scoring
27:13
system,
27:13
I
27:13
imagine.
27:14
So
27:14
to
27:14
bring
27:14
that
27:15
back,
27:15
you
27:15
know,
27:15
I
27:16
understand
27:16
the
27:16
pain
27:17
of
27:17
procurement,
27:18
particularly
27:19
within
27:19
the
27:19
public
27:19
sector.
27:20
Is
27:20
it
27:20
a
27:20
worthwhile
27:21
mission
27:21
and
27:21
worth
27:22
doing?
27:22
Yes,
27:23
absolutely.
27:24
Is
27:24
it
27:24
something
27:25
you
27:25
should
27:25
start
27:25
off
27:25
with?
27:26
No,
27:27
no,
27:27
you
27:27
shouldn't
27:28
avoid.
27:29
So
27:30
there's
27:31
another
27:31
tidbit
27:31
of
27:31
advice
27:32
to
27:32
fellow
27:33
founders.
27:34
Do
27:34
not
27:34
dive
27:34
straight
27:35
in,
27:35
into
27:35
trying
27:36
to
27:36
get
27:36
government
27:36
contracts
27:37
or
27:37
government
27:37
departments
27:38
or
27:38
council
27:38
contracts,
27:39
because
27:39
the
27:40
public
27:40
sector,
27:41
as
27:41
they
27:41
call
27:41
it
27:42
here
27:42
in
27:42
the
27:42
UK,
27:42
or
27:42
the
27:42
State
27:43
Department,
27:43
the
27:43
us,
27:45
everything
27:45
moves
27:45
pretty
27:46
slow
27:47
because
27:47
there's
27:48
so
27:48
many
27:48
stakeholders
27:49
and
27:49
it's
27:49
much
27:49
more
27:50
complicated
27:50
than
27:51
you
27:51
think
27:51
at
27:52
the
27:52
beginning.
27:52
Would
27:52
that
27:52
be
27:53
fair?
27:54
Yeah,
27:54
that's
27:55
very,
27:55
I
27:56
was
27:56
nodding
27:56
away
27:56
there,
27:57
but
27:57
yeah,
27:58
that
27:58
would
27:58
be
27:59
very
27:59
fair.
27:59
And
27:59
look,
28:00
I
28:00
know,
28:01
you
28:01
know,
28:01
we,
28:02
you
28:02
want
28:02
to,
28:03
you
28:03
talk
28:03
a
28:03
lot
28:03
about,
28:03
no,
28:04
no
28:04
BS
28:05
kind
28:05
of
28:05
content
28:06
here.
28:06
So
28:07
just
28:07
to
28:07
be
28:07
clear,
28:07
if,
28:08
if
28:08
you
28:08
are
28:08
thinking
28:09
of
28:09
starting
28:09
with
28:10
the
28:10
public
28:10
sector
28:10
first
28:11
or
28:11
you
28:11
think,
28:11
hey,
28:12
my
28:12
business
28:12
model
28:13
is
28:13
very
28:14
well
28:14
suited
28:14
to
28:14
the
28:14
public
28:15
sector.
28:16
Have
28:17
reserves
28:17
of
28:17
cash.
28:18
Be
28:19
extremely
28:20
intelligent
28:21
with
28:21
your
28:22
burn.
28:22
Very
28:23
conservative,
28:23
Be
28:24
very
28:24
conservative.
28:25
You
28:25
know,
28:26
make
28:26
sure
28:26
you
28:26
have
28:27
contacts,
28:27
warm
28:28
contacts,
28:28
who
28:28
can
28:28
move
28:29
things
28:29
along
28:29
at
28:30
those
28:30
departments.
28:31
Make
28:31
sure
28:31
you
28:31
have
28:32
a
28:32
champ,
28:33
well
28:33
at
28:33
those,
28:33
you
28:33
know,
28:34
particularly
28:34
if
28:34
you're
28:34
up
28:35
against
28:35
some
28:36
well
28:36
known
28:36
competitors
28:37
because
28:38
you
28:38
can
28:38
bet
28:39
they'll
28:39
have
28:39
all
28:39
these
28:39
things
28:40
as
28:41
well.
28:41
So
28:41
it's
28:42
not
28:42
to
28:42
say
28:42
never
28:43
do
28:43
it,
28:43
but
28:44
it's
28:44
very
28:45
rare
28:45
that
28:45
you
28:45
should
28:45
do
28:46
it.
28:46
And
28:46
if
28:46
you
28:46
do,
28:47
you
28:47
should
28:47
be
28:47
extremely
28:48
conservative
28:48
and
28:49
very
28:49
prepared
28:50
to
28:50
wait
28:50
it
28:50
out
28:51
to
28:51
win.
28:52
So
28:52
fundamentally,
28:53
what
28:53
I
28:53
think
28:53
you're
28:54
saying
28:54
is
28:54
when
28:55
it
28:55
comes
28:55
to
28:55
that
28:55
scenario
28:56
planning,
28:56
be
28:56
uber
28:57
conservative.
28:58
And
28:58
yes,
28:59
also,
28:59
to
28:59
use
28:59
a
28:59
weird
29:00
analogy,
29:01
it's
29:01
my
29:01
brain.
29:02
It's
29:02
a
29:02
bit
29:03
like
29:03
going
29:03
into
29:04
a
29:04
sport
29:05
where
29:05
somebody's
29:06
been
29:06
training
29:07
for
29:08
years
29:08
and
29:09
you
29:09
come
29:09
in,
29:09
you
29:09
go,
29:10
I'm
29:10
talented,
29:10
I'm
29:11
new.
29:12
Yeah,
29:12
but
29:12
you
29:12
don't
29:13
have
29:13
any
29:13
of
29:13
the
29:13
experience,
29:14
you
29:14
don't
29:14
have
29:14
any
29:14
of
29:14
the
29:14
knowledge.
29:15
And
29:15
to
29:15
one
29:15
extent
29:16
that
29:16
may
29:16
be
29:16
helpful,
29:16
but
29:17
in
29:17
the
29:17
public
29:17
sector
29:17
it
29:18
really
29:18
isn't
29:18
because
29:18
there's
29:19
so
29:19
many
29:21
hurdles.
29:22
It
29:22
would
29:22
seem
29:23
there
29:23
are.
29:23
There's
29:24
so
29:24
many
29:24
hurdles.
29:24
And
29:25
actually
29:25
I
29:25
say
29:26
that,
29:26
but
29:26
it's
29:26
not
29:27
so
29:27
different
29:27
with
29:28
schools.
29:28
The
29:29
difference
29:29
is
29:30
that
29:30
schools
29:30
will
29:30
have
29:31
a
29:31
certain
29:31
budget,
29:32
particularly
29:32
fee
29:33
paying
29:33
schools,
29:33
so
29:33
private
29:34
slash
29:34
independent.
29:35
But
29:35
these
29:36
schools
29:36
will
29:36
have
29:37
a
29:37
budget
29:37
that
29:38
they
29:38
can
29:38
commit
29:39
to
29:39
for
29:39
learning
29:40
resources.
29:40
They
29:41
will
29:41
have
29:41
a
29:41
threshold
29:42
under
29:42
which
29:43
they
29:43
can
29:43
go
29:43
ahead
29:43
and
29:44
make
29:44
purchases
29:45
without
29:45
needing
29:46
the
29:46
formal
29:46
procurement
29:48
approval.
29:48
And
29:48
so
29:49
that's
29:49
how
29:49
we've
29:49
gone
29:50
forth
29:50
with
29:50
our
29:50
pricing
29:50
strategy.
29:51
And
29:51
look,
29:51
most
29:52
importantly,
29:53
these
29:53
relationships
29:53
are
29:54
built
29:54
on
29:54
trust.
29:55
And
29:56
as
29:56
the
29:56
new
29:56
kid
29:56
on
29:57
the
29:57
block,
29:57
we
29:57
can't
29:58
come
29:58
in
29:58
and
29:58
feasibly
29:59
claim
29:59
that
29:59
we
30:00
are
30:00
trustworthy,
30:01
even
30:01
if
30:01
we
30:01
are
30:01
the
30:01
most
30:02
trustworthy
30:02
people
30:02
on
30:02
the
30:03
planet,
30:03
because
30:03
reputation
30:04
is
30:04
earned
30:05
and
30:05
it's
30:05
earned
30:05
over
30:06
time.
30:06
And
30:07
so
30:08
that's
30:08
why
30:08
we've
30:08
gone
30:08
through
30:09
our
30:09
distribution
30:09
channel.
30:10
So
30:10
we've
30:11
chosen
30:11
a
30:11
very
30:11
specific
30:12
strategy.
30:13
Is
30:13
it,
30:13
is
30:14
it
30:14
a
30:14
great
30:14
industry
30:15
with
30:15
loads
30:16
of
30:16
room
30:16
to
30:16
grow,
30:17
to
30:17
be
30:17
lucrative,
30:18
but
30:18
also
30:18
do
30:18
social
30:19
good?
30:19
So
30:19
this
30:19
sort
30:20
of
30:20
idea
30:20
of
30:20
social
30:21
impact.
30:21
Yes.
30:22
Is
30:23
it
30:23
a
30:23
good
30:23
product?
30:24
Well,
30:24
I'm
30:25
biased,
30:25
I'm
30:25
building
30:25
it.
30:26
So
30:26
I'm
30:26
going
30:26
to
30:26
say
30:26
yes.
30:28
You
30:28
heard
30:28
it
30:28
here
30:28
folks.
30:31
Sorry
30:31
about
30:32
that.
30:32
Shameless
30:32
plug.
30:34
Is
30:34
it
30:35
the
30:35
right
30:36
product
30:38
for
30:39
this
30:40
problem
30:40
we're
30:41
trying
30:41
to
30:41
solve?
30:42
That's
30:42
the
30:43
big
30:43
product
30:43
market
30:44
fit
30:44
question.
30:44
Right.
30:45
And
30:45
that's
30:45
that
30:45
my
30:46
co
30:46
founder
30:46
is
30:46
doing
30:47
a
30:47
lot
30:47
of
30:47
excellent
30:48
work
30:48
going
30:48
out
30:48
there
30:48
and
30:49
looking
30:49
at
30:50
product
30:50
market
30:50
validation.
30:51
We're
30:51
talking
30:52
to
30:52
actual
30:52
schools
30:53
heads,
30:53
governors,
30:54
very
30:55
senior
30:55
people
30:56
and
30:56
they
30:56
like
30:56
what
30:57
they're
30:57
seeing.
30:57
So
30:58
yeah,
30:58
it's
30:58
not,
30:59
not
30:59
blindly
31:00
that
31:00
we've
31:00
turned
31:00
away
31:00
from
31:01
enterprise,
31:01
but
31:02
it's,
31:02
it's
31:02
a
31:02
calculated
31:03
risk.
31:03
It's
31:04
all
31:04
risk
31:04
at
31:04
the
31:04
end
31:04
of
31:04
the
31:05
day.
31:05
But
31:05
it's,
31:05
it's
31:05
a
31:06
calculated
31:06
risk.
31:07
It's,
31:07
it's
31:07
a
31:07
really.
31:08
I,
31:08
I
31:08
think
31:09
what
31:09
you're,
31:09
what
31:10
you're
31:10
trying
31:10
to
31:10
achieve
31:11
is,
31:11
is,
31:12
is
31:12
really
31:12
great
31:13
because
31:14
ultimately
31:15
you
31:15
know,
31:15
AI
31:16
is,
31:16
is
31:17
coming,
31:17
growing,
31:18
developing,
31:18
going
31:19
to
31:19
be,
31:19
be
31:19
you
31:20
know,
31:20
the
31:20
people
31:20
who
31:20
say
31:21
I'm
31:21
never
31:21
going
31:21
to
31:21
touch
31:22
it.
31:22
I
31:22
think
31:24
it's
31:24
noble
31:24
to
31:25
think
31:25
that
31:25
like
31:27
I
31:27
don't
31:27
know,
31:27
I
31:27
don't
31:27
know
31:27
where
31:28
I
31:28
guess,
31:29
you
31:29
know,
31:29
human
31:30
beings,
31:30
I
31:30
could
31:31
go
31:31
into
31:31
this
31:31
and
31:32
for
31:32
hours
31:32
but
31:33
change.
31:33
Right.
31:33
And
31:34
it's
31:34
like
31:34
the
31:34
organizations
31:35
that
31:35
will
31:36
probably
31:37
worry
31:37
the
31:37
most,
31:38
I
31:38
would
31:38
say
31:38
are
31:39
probably
31:39
the
31:40
businesses
31:40
that
31:40
are
31:40
smaller
31:41
than
31:41
your
31:41
49
31:42
to
31:42
5
31:42
hundreds,
31:43
your,
31:43
your
31:44
1
31:44
to
31:44
50s,
31:44
you
31:45
know,
31:45
where
31:45
how
31:45
do
31:46
we
31:46
make
31:46
efficiencies?
31:46
And
31:47
actually
31:47
the
31:47
gains
31:47
can
31:47
be
31:48
most
31:48
significant
31:48
because
31:49
in
31:50
a
31:50
way,
31:50
you
31:51
know,
31:51
if
31:51
you're
31:51
only
31:51
a
31:51
team
31:52
of
31:52
10
31:53
and
31:53
you're
31:53
suddenly
31:54
able
31:54
to
31:54
triple
31:55
productivity.
31:55
Wow.
31:56
It's
31:57
much
31:57
more
31:57
difficult
31:57
to
31:57
do
31:57
that
31:58
in
31:58
a
31:58
huge,
31:58
huge
31:59
company.
31:59
It
32:00
is,
32:00
but
32:00
also
32:01
mid
32:01
size.
32:02
So
32:02
without
32:03
I
32:04
obviously
32:04
I
32:05
can't
32:05
say
32:05
too
32:05
much
32:05
about
32:06
my
32:09
next
32:10
one
32:10
I
32:11
think
32:12
well
32:12
publicized
32:13
problem
32:14
that
32:14
other
32:14
companies
32:15
are
32:15
having.
32:15
They're
32:16
also
32:16
having.
32:16
So
32:17
Colombiax
32:18
was
32:18
born
32:18
in,
32:20
I
32:20
believe
32:20
it
32:20
was
32:21
20,
32:22
early
32:22
20,
32:23
21
32:23
or
32:23
so.
32:24
So
32:24
again
32:25
that
32:25
era
32:25
of
32:25
cheap
32:25
money.
32:26
Right.
32:26
After
32:26
Covid
32:27
did,
32:27
restrictions
32:28
lifted
32:28
money
32:28
flow
32:29
back
32:29
in
32:29
into
32:30
the
32:30
stock
32:30
markets.
32:31
Interest
32:31
rates
32:32
were
32:32
still
32:32
at
32:33
the
32:33
base
32:33
rate
32:33
or
32:33
so
32:34
of
32:34
central
32:34
banks.
32:34
And
32:35
so,
32:35
yeah,
32:35
investors
32:36
needed
32:36
somewhere
32:36
to
32:36
park
32:37
their
32:37
money.
32:38
And
32:38
so
32:38
great
32:38
time
32:39
for,
32:39
you
32:39
know,
32:39
a
32:39
lot
32:39
of
32:40
money.
32:40
I
32:40
think
32:40
that
32:40
was
32:40
a
32:40
record
32:41
year,
32:41
if
32:41
I'm
32:42
not
32:42
wrong.
32:42
I
32:42
don't
32:42
want
32:42
to
32:43
misspeak,
32:43
but
32:43
for
32:43
private
32:44
investment
32:44
into
32:45
tech
32:45
companies
32:46
effectively
32:46
as
32:46
a
32:47
whole,
32:47
and
32:47
particularly
32:47
software
32:48
companies,
32:48
I
32:49
think
32:49
it's
32:49
fair
32:49
to
32:49
say
32:50
that,
32:50
that
32:50
those
32:50
companies
32:51
were
32:51
both
32:51
fortunate
32:52
and
32:52
very
32:52
and
32:53
unfortunate.
32:53
They
32:54
were
32:54
fortunate
32:54
because
32:55
they
32:55
received
32:55
plentiful
32:56
capital,
32:56
they
32:57
were
32:57
able
32:57
to
32:57
get
32:57
off
32:57
the
32:57
ground
32:58
quickly.
32:58
They
32:58
faced
32:59
no
32:59
sort
33:00
of
33:00
enormous,
33:01
terrible
33:01
macroeconomic
33:02
headwinds
33:03
because
33:03
we'd
33:03
just
33:03
been
33:04
through
33:04
a
33:04
couple
33:05
of
33:05
a
33:05
year
33:05
or
33:05
so
33:05
of
33:06
once
33:06
in
33:07
a
33:07
lifetime
33:07
pandemic
33:08
and
33:08
we
33:08
were
33:08
at
33:09
the
33:09
point
33:09
now,
33:09
you
33:09
know,
33:10
where,
33:10
where
33:10
they
33:10
could
33:10
just
33:10
launch
33:11
and
33:11
start
33:11
hiring
33:11
people.
33:12
And
33:12
that
33:12
old
33:12
mentality,
33:13
sorry
33:13
I
33:13
say
33:14
old,
33:14
but
33:14
the
33:14
mentality
33:15
of
33:15
the
33:15
past,
33:15
like
33:15
five
33:16
to
33:16
eight
33:16
years
33:16
was
33:16
still
33:17
there,
33:17
right?
33:17
So
33:17
it
33:18
was
33:18
all
33:18
that
33:18
money.
33:18
Don't
33:19
worry
33:19
about
33:19
making
33:19
a
33:19
profit
33:20
right
33:20
now.
33:20
Take
33:20
the
33:21
investor
33:21
money,
33:22
burn
33:22
that
33:22
cash,
33:23
get
33:24
your
33:24
users,
33:25
get
33:25
them
33:25
to
33:25
stay
33:26
on
33:26
the
33:26
app
33:27
while
33:27
you
33:27
acquire
33:27
new
33:27
users,
33:28
just
33:28
become
33:28
big.
33:29
Don't
33:29
worry
33:29
about
33:29
the
33:29
fundamentals
33:30
of
33:30
businesses.
33:31
Right.
33:31
Obviously
33:32
that's,
33:32
that's
33:33
very
33:33
different
33:33
now,
33:34
but,
33:34
but
33:34
at
33:34
that
33:34
time
33:35
that,
33:35
that's
33:35
what
33:35
it
33:35
was
33:35
and
33:36
that's
33:36
what
33:36
it
33:36
felt
33:37
like,
33:37
you
33:37
know,
33:37
particularly
33:37
for
33:38
those
33:38
of
33:38
us
33:38
who
33:38
had
33:38
our
33:38
CVs
33:39
out
33:39
in
33:39
the
33:39
job
33:40
market
33:40
and
33:40
the
33:41
year
33:41
after.
33:42
So,
33:42
so
33:42
they'd
33:43
built
33:43
up
33:43
all
33:43
this
33:43
infrastructure,
33:44
they'd
33:44
hired
33:44
all
33:44
these
33:45
people.
33:45
The
33:45
year
33:45
after
33:46
2022,
33:47
ChatGPT
33:48
was
33:48
released.
33:48
And
33:48
it's
33:49
a
33:49
technology
33:50
GPT.
33:50
The
33:51
T
33:51
stands
33:51
for
33:52
transformer,
33:52
which
33:52
is
33:53
the
33:53
type
33:53
of
33:53
model
33:54
they're
33:54
using
33:54
for
33:54
the
33:54
large
33:54
language
33:55
model.
33:55
And
33:55
the
33:56
key
33:56
difference
33:56
here
33:56
is
33:58
this
33:58
structure
33:59
of
33:59
AI
33:59
was
34:00
developed
34:00
in
34:00
a
34:00
Google
34:01
Brain
34:01
paper
34:01
that
34:02
was
34:02
published,
34:03
but
34:03
it
34:03
was
34:03
commercialized
34:05
by,
34:05
and
34:06
it
34:06
was
34:06
released
34:07
by
34:07
Sam
34:08
Altman,
34:08
who's
34:08
very
34:09
much
34:09
sort
34:09
of
34:09
a
34:09
product
34:10
person.
34:10
And
34:10
to
34:11
their
34:11
great
34:11
surprise,
34:12
it
34:12
took
34:12
off.
34:12
And
34:13
so
34:13
the
34:13
applications
34:14
came
34:14
very
34:14
soon
34:15
after
34:15
that.
34:16
Initially
34:16
it
34:16
was
34:16
coding
34:17
completion.
34:17
So
34:17
if
34:18
we
34:18
talk
34:18
about
34:18
software
34:19
here,
34:19
specifically
34:20
it
34:20
was
34:20
coding
34:20
completion
34:21
with
34:21
Microsoft's
34:22
Copilot.
34:22
And
34:22
then
34:23
there
34:23
were
34:23
a
34:23
variety
34:23
of
34:24
competing
34:24
products.
34:25
It
34:25
was
34:25
all
34:25
about
34:25
helping
34:26
developers
34:26
build
34:27
little
34:27
bit
34:27
at
34:28
a
34:28
time
34:28
with
34:28
a
34:28
bit
34:28
more
34:29
assistance
34:29
than
34:29
they
34:30
were
34:30
used
34:30
to.
34:30
And
34:30
then
34:31
we
34:31
moved
34:31
into,
34:32
you
34:32
know,
34:32
a
34:32
few
34:32
years
34:32
later,
34:33
it
34:33
kept
34:33
developing
34:34
and
34:34
it
34:34
kept
34:34
getting
34:34
better.
34:35
But
34:35
then
34:35
we,
34:35
we
34:35
got
34:36
cursor,
34:36
we
34:37
got
34:37
Claude
34:37
code
34:38
not
34:38
to
34:38
Promote
34:38
certain
34:39
providers
34:41
lovable.
34:42
The
34:42
whole.
34:43
Yeah,
34:43
basically.
34:44
Exactly.
34:45
And,
34:45
and
34:45
so
34:45
now,
34:46
you
34:46
know,
34:46
we,
34:48
if
34:48
you
34:48
look
34:48
at
34:48
this
34:48
from
34:48
a
34:49
business
34:50
point
34:50
of
34:51
view
34:51
that
34:51
started
34:51
up
34:51
in
34:52
early
34:52
2021,
34:53
you're
34:53
now
34:53
at
34:53
a
34:53
point
34:53
where
34:54
you've
34:54
expanded
34:54
everything,
34:55
you've
34:55
burnt
34:55
this
34:55
load
34:55
of
34:56
cash,
34:56
you've
34:56
built
34:57
this
34:57
reputation
34:58
of
34:58
like,
34:58
we've
34:58
got
34:59
the
34:59
best
34:59
engineers,
35:00
we
35:00
build
35:00
this,
35:00
we
35:01
do
35:01
this,
35:01
we
35:02
do
35:02
it
35:02
all
35:02
by
35:02
the
35:03
ways
35:05
of
35:05
agile
35:06
product
35:06
development
35:06
that
35:07
have
35:07
been
35:07
set
35:07
over
35:07
the
35:07
past
35:08
few
35:08
years.
35:09
Now
35:10
all
35:10
of
35:10
a
35:10
sudden,
35:10
oh,
35:11
you
35:11
know,
35:11
engineers
35:11
are
35:12
taking
35:12
quite
35:12
a
35:12
long
35:12
time
35:13
to
35:13
build
35:13
something.
35:13
Why
35:14
don't
35:14
they
35:14
just
35:14
ship
35:14
it
35:14
with,
35:15
with,
35:15
with
35:15
AI,
35:15
you
35:16
know,
35:16
you've
35:16
got
35:16
investors
35:17
who
35:17
are
35:17
not
35:18
particularly
35:18
VCs,
35:19
who
35:19
with
35:19
all
35:19
due
35:20
respect,
35:20
are
35:20
not
35:21
always
35:21
the
35:21
most
35:21
technically
35:22
savvy
35:22
people
35:23
and
35:23
you've
35:24
kind
35:24
of
35:24
got
35:24
them
35:24
putting
35:25
pressure.
35:25
Right.
35:26
So
35:26
they're
35:26
saying,
35:26
wow,
35:26
look
35:27
at
35:27
this
35:27
business
35:27
that
35:28
does
35:28
something
35:28
very
35:28
similar
35:29
to
35:29
what
35:29
you
35:29
do.
35:30
You
35:31
know,
35:31
it's,
35:31
it's
35:31
launched
35:32
two
35:32
years
35:33
after
35:33
you
35:33
and
35:33
it's
35:34
taken
35:34
more
35:34
of
35:34
an
35:34
AI
35:34
native
35:35
approach
35:35
and
35:35
it's
35:36
producing
35:36
more,
35:37
more
35:37
quickly
35:37
than
35:37
you.
35:37
So
35:38
the
35:38
velocity
35:38
is,
35:38
is
35:39
that
35:39
it's
35:39
got
35:40
a
35:40
far
35:40
smaller
35:40
headcount
35:41
as
35:41
a
35:41
result,
35:42
it's
35:42
far
35:42
more
35:42
profitable.
35:43
And
35:43
so
35:44
why
35:44
aren't
35:44
you
35:44
doing
35:44
that?
35:45
And
35:45
so
35:45
this
35:45
was
35:46
a
35:46
very
35:46
real
35:46
problem
35:47
at
35:47
the
35:47
time
35:47
I
35:47
was
35:48
leaving
35:48
Climatex.
35:49
But
35:49
it's
35:49
not
35:49
specific
35:50
to
35:50
Climax.
35:50
You
35:50
know,
35:51
it's
35:51
a
35:51
group
35:51
of
35:52
companies
35:52
there.
35:53
It's
35:53
rocked
35:53
a
35:53
lot
35:54
of
35:54
software
35:54
companies,
35:54
hasn't
35:55
it?
35:55
Yeah.
35:56
On
35:57
the
35:57
stock
35:57
market.
35:57
Yeah.
35:58
So
35:59
like
35:59
Salesforce,
35:59
HubSpot,
36:00
blah,
36:01
blah,
36:01
blah,
36:01
blah,
36:01
blah.
36:01
Interestingly,
36:02
like,
36:03
when
36:03
it
36:03
comes
36:03
to
36:04
just
36:04
taking
36:04
your
36:05
point,
36:05
so
36:05
you
36:05
obviously
36:06
worked
36:06
on
36:06
building
36:07
LLMs.
36:08
You're
36:08
looking
36:09
to
36:09
fuel
36:10
the
36:10
next
36:10
generation
36:12
with
36:12
AI
36:13
knowledge
36:13
and
36:13
expertise
36:14
and
36:14
really
36:14
create
36:14
that
36:15
benchmark.
36:15
Now,
36:16
you
36:16
touched
36:16
on
36:16
Copilot.
36:17
Now
36:17
I
36:18
used
36:18
a
36:18
lot
36:18
of
36:18
different
36:19
models.
36:20
I'm
36:20
not
36:20
saying
36:20
I'm
36:20
anywhere
36:21
near
36:21
your
36:21
level.
36:22
You're,
36:22
you
36:22
know,
36:23
probably
36:24
much
36:24
more
36:25
effective
36:25
than
36:26
me
36:26
personally,
36:26
but
36:27
certainly
36:27
when
36:27
I
36:27
talk
36:28
with
36:28
founders
36:28
and
36:29
when
36:29
I
36:29
play
36:30
with
36:30
them
36:30
myself,
36:31
it
36:31
seems
36:31
like
36:31
Copilot
36:33
seems
36:33
to
36:33
have
36:34
fallen
36:34
considerably
36:35
behind
36:35
in
36:35
the
36:36
race
36:36
and,
36:37
and
36:37
like
36:37
they,
36:38
I
36:38
saw
36:38
they
36:39
were
36:39
only
36:39
able
36:39
to
36:39
sign
36:40
up
36:40
3%
36:40
of
36:41
people
36:41
onto
36:41
a
36:41
paid
36:42
plan.
36:42
It's
36:42
an
36:43
interesting
36:43
thing.
36:43
Right,
36:44
so
36:44
like
36:44
taking,
36:45
if
36:45
we
36:46
talk
36:46
about.
36:46
And
36:47
we'll
36:47
see
36:47
how
36:47
well
36:47
this
36:47
ages,
36:48
mate,
36:48
I'll
36:48
have
36:48
you
36:49
back
36:49
in
36:49
a
36:49
couple
36:49
of
36:49
years
36:49
and
36:50
we'll
36:50
see
36:50
whether
36:50
we're
36:50
still
36:51
talking
36:51
about
36:51
Claude
36:52
chatgpt,
36:53
Copilot,
36:55
et
36:55
cetera,
36:56
in
36:56
terms
36:56
of
36:56
preparing
36:57
young
36:57
People
36:57
for
36:58
the
36:58
future,
36:59
which
36:59
is
36:59
your
37:00
goal.
37:02
What
37:02
are
37:02
you
37:02
doing
37:03
in
37:03
terms
37:03
of.
37:04
Are
37:04
you
37:04
trying
37:04
to.
37:05
And
37:06
bear
37:06
in
37:06
mind
37:06
people
37:06
listening.
37:07
I
37:07
haven't
37:08
been
37:08
through
37:08
all
37:10
Atish's
37:11
courses,
37:11
so
37:12
I'm
37:13
not
37:13
going
37:13
to
37:13
do
37:13
a
37:13
shameless
37:14
plug
37:14
here.
37:15
But
37:15
ultimately,
37:16
are
37:17
you
37:17
preparing
37:17
them
37:18
for
37:18
the
37:18
different
37:18
models?
37:19
Are
37:19
you
37:19
preparing
37:20
them?
37:20
Like,
37:21
how
37:21
are
37:21
you
37:21
thinking
37:21
through
37:22
this
37:22
for
37:22
the
37:22
future?
37:23
We're
37:23
sort
37:23
of.
37:24
We
37:24
teach
37:24
model
37:24
agnostic.
37:26
There
37:26
are
37:27
effectively
37:27
what
37:28
we
37:28
do
37:28
in,
37:28
in
37:28
the
37:28
product
37:29
is
37:29
we
37:30
run
37:30
what
37:30
we
37:30
call
37:30
pulse
37:31
updates.
37:31
So
37:32
that's.
37:32
They
37:32
run
37:33
every
37:33
sort
37:33
of
37:34
quarter
37:34
and
37:34
they
37:34
keep
37:34
content
37:35
fresh
37:35
because
37:36
this
37:36
is
37:36
a
37:36
field
37:36
that's
37:36
moving
37:37
at
37:37
breakneck
37:38
speed
37:38
so
37:38
fast.
37:38
And
37:39
so
37:39
the
37:39
idea
37:39
is
37:39
people
37:40
go
37:40
in,
37:40
they
37:40
get
37:41
their
37:41
certificate,
37:42
right?
37:42
So
37:42
we'll,
37:43
we'll
37:43
be
37:43
sort
37:43
of
37:44
CPD
37:44
certified,
37:45
which
37:45
is
37:45
like
37:45
a
37:46
form,
37:46
at
37:47
least
37:47
in
37:47
the
37:47
uk,
37:47
of
37:48
recognized
37:48
in
37:49
education,
37:49
continuing
37:50
sort
37:50
of
37:50
professional
37:51
development.
37:51
Once
37:52
they've
37:52
got
37:52
that
37:53
certificate,
37:53
it's
37:53
locked
37:54
for
37:54
a
37:54
year
37:54
because
37:54
teachers,
37:55
educators
37:56
are
37:56
busy
37:56
enough.
37:56
You
37:56
don't
37:57
want
37:57
them
37:57
to
37:57
go
37:57
and
37:57
refresh
37:58
contact
37:58
every
37:58
month,
37:59
every
37:59
few
37:59
months
37:59
or
38:00
every
38:00
month.
38:00
It's
38:01
too
38:01
much.
38:01
You're
38:01
then
38:01
adding
38:02
work
38:02
and
38:02
then
38:02
it's.
38:03
You're
38:03
thinking,
38:03
well,
38:03
like,
38:04
are
38:04
we
38:05
really
38:05
solving
38:05
their
38:05
problem
38:05
or
38:06
are
38:06
we
38:06
just
38:06
adding
38:07
more
38:07
of
38:07
a
38:07
burden?
38:07
But
38:08
so
38:08
we're
38:09
refreshing
38:09
a
38:09
lot
38:09
of
38:09
that
38:10
content.
38:10
We
38:10
are
38:10
model
38:11
agnostic
38:11
and
38:12
we're
38:12
focused
38:13
a
38:13
lot
38:13
at
38:14
the
38:14
moment
38:14
on
38:15
generative
38:15
AI.
38:16
And
38:17
this
38:18
is
38:18
a
38:18
deliberate
38:18
choice.
38:20
You'll
38:20
often
38:21
hear
38:21
that
38:22
the,
38:22
for
38:22
example,
38:23
that
38:23
the
38:23
U.S.
38:24
army
38:24
or
38:25
the
38:25
U.S.
38:25
military
38:26
or
38:26
a
38:27
lab
38:27
somewhere
38:27
or
38:28
whatever
38:29
it
38:29
be,
38:30
has
38:30
used
38:31
AI,
38:32
just
38:32
those
38:32
letters,
38:33
you'll
38:33
see,
38:33
to
38:34
accomplish
38:34
something
38:35
fantastic.
38:36
And
38:36
then
38:36
it's
38:37
not
38:37
until
38:37
you
38:38
click
38:38
through
38:38
the
38:39
headline
38:39
and
38:39
you
38:40
read
38:40
the
38:40
article
38:40
that
38:40
you
38:41
understand.
38:41
No,
38:42
they've
38:42
kind
38:42
of
38:42
built
38:42
their
38:42
own
38:43
model.
38:43
So
38:43
it
38:43
might
38:44
be
38:44
more
38:44
traditional
38:45
AI,
38:45
it
38:46
might
38:46
be,
38:47
sorry,
38:47
by
38:47
which
38:47
I
38:47
mean
38:48
sort
38:48
of
38:48
algorithmic
38:49
based.
38:49
It
38:49
might
38:50
be
38:50
that
38:50
they've
38:50
taken
38:50
their
38:50
own
38:51
generative
38:51
AI
38:51
model
38:52
and
38:52
done
38:52
something
38:53
totally
38:53
different
38:54
with
38:54
it
38:55
and
38:55
so
38:55
on
38:55
and
38:55
so
38:55
on.
38:56
But
38:56
we,
38:56
we're
38:57
focusing
38:57
on
38:58
impact,
38:59
right?
38:59
We're
38:59
really
39:00
focused
39:00
on
39:00
impact.
39:01
So
39:02
what
39:02
can
39:03
educators
39:03
pick
39:03
up
39:04
today
39:05
that
39:05
after
39:06
a
39:07
few
39:07
days
39:08
worth
39:08
of
39:08
courses
39:09
or
39:09
a
39:09
few
39:09
hours
39:09
worth
39:10
of
39:10
courses
39:10
in
39:10
that
39:11
micro
39:12
learning
39:12
element,
39:12
so
39:13
that
39:13
gamified
39:13
element,
39:14
what
39:15
can
39:15
they
39:16
go
39:16
from
39:17
again,
39:17
0
39:17
to
39:18
1
39:18
and
39:18
where
39:19
can
39:19
they
39:19
immediately
39:20
start
39:20
applying
39:21
it
39:21
and
39:21
where
39:21
can
39:21
they
39:22
have
39:22
impact?
39:22
So
39:23
we're
39:23
model
39:23
agnostic.
39:24
We're
39:24
focusing
39:24
on
39:25
the
39:25
cutting
39:25
edge
39:25
of
39:26
AI
39:26
at
39:26
the
39:27
moment,
39:27
which
39:27
is
39:27
generative
39:28
AI,
39:28
specifically
39:29
through
39:29
large
39:29
language
39:30
models.
39:31
And
39:31
yes,
39:31
as
39:32
they
39:32
are
39:32
applied
39:33
through
39:33
tools.
39:35
So
39:35
we
39:35
talk
39:35
about
39:35
various
39:36
tool
39:36
examples,
39:37
but
39:38
the
39:39
idea
39:39
is
39:39
not
39:39
to
39:39
train
39:40
people
39:40
on
39:40
specific
39:40
technologies
39:41
because
39:42
we
39:42
don't
39:42
have
39:42
those
39:42
alliances
39:43
ourselves.
39:44
And
39:44
I
39:44
think
39:44
if
39:44
we
39:45
did,
39:45
it
39:45
would
39:45
skew
39:45
our
39:46
approach
39:46
and
39:46
we
39:46
wouldn't
39:46
be
39:47
as
39:47
trustful.
39:48
You
39:48
might
39:48
be
39:48
pulled
39:49
in
39:49
a
39:49
certain
39:50
direction,
39:50
it
39:51
might
39:51
be
39:51
biased
39:52
rather
39:52
than.
39:53
Yeah,
39:53
yeah,
39:53
that's
39:53
great.
39:54
Exactly.
39:55
And
39:55
so,
39:56
yeah,
39:56
we're
39:57
very
39:57
much
39:57
focused
39:58
on
39:58
real
39:58
world
39:58
impact
39:59
tools
39:59
that
39:59
are
39:59
out
40:00
now.
40:00
We're
40:00
refreshing
40:01
this
40:01
content.
40:01
If
40:02
you
40:02
open
40:02
up,
40:02
start,
40:03
I
40:03
don't
40:03
know,
40:03
Chat,
40:03
GPT
40:04
or
40:04
Claude
40:04
or
40:05
Gemini
40:05
or
40:05
Grok
40:05
or
40:06
what
40:06
have
40:06
you
40:06
and
40:06
others
40:07
that
40:07
I
40:07
haven't
40:07
mentioned.
40:07
And
40:08
you
40:08
go
40:08
through
40:08
the
40:09
Chatbot
40:09
interface,
40:09
what
40:10
can
40:10
you
40:10
accomplish
40:10
there
40:11
versus
40:11
accomplish
40:12
using
40:12
some
40:12
of
40:12
the
40:13
tools
40:13
that,
40:13
that
40:13
are
40:14
out
40:14
there.
40:14
And
40:15
I
40:15
think
40:15
in
40:15
the
40:15
longer
40:15
term,
40:16
like,
40:16
we're
40:16
also
40:16
thinking
40:16
about
40:17
how
40:17
can
40:17
we
40:17
bring
40:17
some
40:18
of
40:18
those
40:18
tools
40:19
into
40:20
an
40:20
educator's
40:21
work?
40:22
So
40:22
they're
40:23
different
40:23
work.
40:24
Because
40:24
they
40:25
are,
40:25
they
40:25
are.
40:26
It's.
40:27
That's
40:27
it.
40:28
Yeah,
40:28
yeah,
40:28
yeah,
40:29
exactly.
40:29
So
40:29
that,
40:30
that's
40:30
sort
40:30
of
40:30
where
40:30
we
40:30
are
40:31
at
40:31
the
40:31
moment.
40:31
If
40:31
I
40:32
may.
40:32
Things
40:32
like
40:33
cardboard
40:34
and
40:35
which
40:35
again,
40:36
this
40:36
may
40:36
date
40:36
the
40:36
podcast
40:37
episode,
40:37
because
40:38
I
40:38
don't
40:38
know
40:38
what's
40:38
going
40:38
to
40:38
happen
40:39
in
40:39
the
40:39
future,
40:39
but
40:40
obviously
40:40
the
40:40
idea
40:41
of
40:41
cloudbot
40:41
was
40:42
building
40:42
an
40:42
agent,
40:43
an
40:43
agent
40:43
that
40:43
interacted
40:44
with
40:44
different
40:45
things.
40:46
Are
40:46
you
40:46
trying
40:47
to
40:47
help
40:47
them
40:48
from
40:48
the
40:48
perspective
40:49
of
40:49
like,
40:49
when
40:49
you
40:50
talk
40:50
about
40:50
LLMs
40:51
and
40:51
general
40:51
skills,
40:52
are
40:52
you
40:52
also
40:53
trying
40:53
to
40:53
prepare
40:53
them
40:54
for
40:54
a
40:54
world
40:54
where
40:54
agents
40:56
will
40:56
be
40:57
trained,
40:57
built,
40:58
developed
40:58
to
40:59
do,
41:00
you
41:00
know,
41:00
task
41:01
based
41:01
stuff?
41:03
Yeah,
41:03
yeah,
41:03
we,
41:04
we
41:04
are.
41:04
So
41:05
in
41:05
terms
41:06
of
41:06
the
41:06
product
41:06
itself,
41:07
the
41:07
first
41:07
part
41:07
is
41:07
sort
41:08
of
41:08
a
41:08
learning
41:08
platform.
41:09
So
41:09
if
41:09
I
41:10
start
41:10
at
41:10
the
41:10
top,
41:10
the
41:11
whole
41:11
point
41:11
of
41:11
pedagogue,
41:12
if,
41:12
if
41:12
we
41:13
were
41:13
to
41:13
put
41:13
it
41:13
in
41:13
a
41:13
sentence
41:14
or
41:14
two,
41:15
is
41:15
to
41:16
set
41:16
a
41:16
standard
41:17
that
41:17
doesn't
41:17
yet
41:17
exist
41:18
about
41:18
how
41:18
AI
41:19
should
41:19
be
41:20
used
41:20
and
41:21
taught,
41:21
taught
41:21
in
41:22
education
41:22
to
41:23
progress
41:23
that
41:24
standard
41:24
over
41:24
time
41:25
and
41:25
keep
41:25
it
41:25
up
41:26
to
41:26
date.
41:27
And
41:27
then
41:27
to
41:27
ensure
41:28
that
41:28
they,
41:28
the
41:29
set
41:29
of
41:30
shared
41:31
standards,
41:31
auditable
41:32
standards,
41:33
are
41:33
effectively
41:34
a
41:35
sort
41:35
of
41:35
new
41:35
baseline
41:36
for
41:37
what
41:37
people
41:37
should
41:38
look
41:38
at.
41:38
The
41:38
bare
41:39
minimum
41:39
they
41:39
should
41:39
look
41:39
at.
41:40
And
41:40
then
41:40
we'll
41:40
structure
41:41
up
41:41
from
41:41
that.
41:41
So
41:41
what's
41:41
the
41:42
bare
41:42
minimum?
41:42
What's
41:42
good
41:43
to
41:43
have,
41:43
what
41:43
could
41:44
you
41:44
have,
41:44
what
41:44
would
41:45
show
41:45
excellence
41:45
and
41:46
so
41:46
on.
41:46
But
41:46
right
41:46
now
41:47
it's
41:47
trying
41:47
to
41:47
create
41:48
that
41:48
baseline,
41:48
that
41:48
common
41:49
baseline
41:49
that's
41:50
really
41:50
key.
41:50
And
41:51
so,
41:51
so
41:52
in
41:52
order
41:53
to
41:53
do
41:53
that,
41:53
we've
41:54
got
41:54
a
41:54
product
41:54
that
41:55
teaches
41:55
people
41:56
using
41:57
again.
41:57
So
41:57
we
41:58
create
41:58
courses,
41:58
we
41:58
keep
41:59
them
41:59
fresh.
42:00
Everything
42:00
I
42:00
was
42:00
Just
42:01
sort
42:01
of
42:01
talking
42:01
about
42:01
there
42:02
applied
42:02
AI.
42:03
How
42:03
can
42:03
it
42:03
be
42:04
most
42:04
impactful?
42:05
We're
42:05
introducing
42:05
a
42:06
sort
42:06
of
42:06
playground
42:07
which
42:07
will
42:08
allow
42:08
people
42:08
to
42:08
practice
42:09
in
42:09
a
42:09
safe
42:09
and
42:09
secure
42:10
environment
42:11
and
42:11
not
42:11
worry
42:12
about
42:12
their
42:12
data
42:13
going
42:13
off
42:14
into
42:14
the
42:15
servers
42:15
somewhere
42:16
else
42:16
or
42:17
whatever
42:17
answers
42:18
they're
42:18
providing.
42:18
Even
42:19
if
42:19
there's
42:19
terms
42:19
and
42:20
conditions
42:20
and
42:20
a
42:20
privacy
42:21
policy
42:21
and
42:21
all
42:21
that
42:22
good
42:22
stuff.
42:22
The
42:22
truth
42:23
is
42:23
that
42:23
this
42:24
is
42:24
a
42:24
legal
42:25
agreement
42:25
and
42:25
set
42:26
against
42:26
what's
42:26
technically
42:27
happening
42:28
on
42:28
the
42:28
back
42:28
end
42:28
of
42:29
any
42:29
platform.
42:29
You
42:29
don't
42:30
know.
42:30
So
42:31
not,
42:31
not
42:31
trying
42:31
to
42:32
be
42:32
cynical
42:32
but
42:32
that's,
42:33
that's
42:33
the
42:33
hard
42:34
truth
42:34
of
42:34
it.
42:34
Well,
42:34
they
42:34
are
42:35
in
42:35
that
42:35
early
42:35
stage
42:35
of
42:36
just
42:36
trying
42:36
to
42:36
gather
42:36
as
42:37
much
42:37
information
42:37
as
42:38
they
42:38
can
42:38
and
42:39
they're
42:39
going
42:39
to
42:39
spin
42:39
off
42:39
products.
42:40
It's
42:40
like
42:40
the
42:40
ultimate
42:41
user
42:42
feedback
42:42
tool.
42:43
Yes.
42:44
That
42:44
everybody's
42:45
using.
42:45
We're
42:46
all
42:46
guinea
42:46
pigs
42:46
basically.
42:48
That's
42:48
it.
42:48
And
42:48
to
42:49
come
42:49
back
42:49
to
42:49
your
42:49
Claude.
42:50
So
42:50
I
42:50
think
42:50
it's
42:50
called
42:50
OpenClaw
42:51
now.
42:51
I
42:52
think
42:52
it
42:52
went
42:52
through
42:52
two
42:53
name
42:53
changes
42:53
or
42:54
something.
42:54
You're
42:54
referring
42:55
to
42:55
the
42:55
multi
42:55
agent
42:56
crazy.
42:57
Yeah,
42:58
the
42:58
one
42:58
that
42:59
everybody
42:59
was
42:59
buying
42:59
Mac
43:00
Minis.
43:00
Oh
43:01
yeah.
43:01
God,
43:01
right.
43:01
That's
43:02
a
43:02
whole
43:02
separate
43:03
conversation
43:03
which
43:03
they
43:04
don't
43:04
do
43:04
that.
43:04
If
43:05
you're
43:05
thinking
43:05
of
43:05
buying
43:05
a
43:05
Mac
43:06
Mini
43:06
just
43:06
to
43:06
run
43:07
it,
43:07
don't
43:07
do
43:07
it.
43:07
There's
43:07
far
43:07
more
43:08
cost
43:08
effects.
43:08
You
43:09
can
43:09
do
43:09
it
43:09
in
43:09
a
43:10
simple
43:10
way,
43:10
I
43:10
imagine.
43:11
And
43:11
leaned
43:11
in
43:11
a
43:12
simpler
43:12
way.
43:12
God,
43:12
yeah,
43:13
that
43:13
was
43:13
fun
43:13
time.
43:15
Good
43:15
time
43:15
for
43:15
Apple
43:16
though.
43:18
And
43:18
the
43:18
secondhand
43:19
market
43:19
I
43:19
imagine.
43:20
And
43:20
the
43:20
secondhand
43:21
market.
43:21
Yeah,
43:21
very
43:22
true.
43:22
So,
43:22
so
43:23
look,
43:23
I,
43:23
I
43:23
think
43:23
there's
43:24
a
43:24
lot
43:24
of
43:24
hype
43:24
about,
43:25
about
43:25
openclaw
43:26
as
43:26
it's
43:26
now
43:26
called
43:26
and
43:27
I
43:27
don't
43:27
want
43:27
to,
43:27
I'm
43:28
not
43:28
gonna
43:29
be
43:29
that
43:29
verbose
43:30
here
43:30
talking
43:31
about
43:31
it.
43:31
What,
43:32
what
43:32
is
43:32
openclaw?
43:33
It's
43:33
actually
43:33
a
43:34
layer
43:35
built
43:35
on
43:35
top
43:36
of
43:36
the,
43:37
you
43:37
know,
43:37
the
43:37
Frontier
43:38
models
43:38
products.
43:38
So
43:39
the,
43:39
the
43:39
large
43:39
language
43:40
models
43:40
and,
43:40
and
43:41
you.
43:42
Or
43:42
several
43:43
API
43:43
keys
43:44
in
43:44
for
43:44
a
43:45
Frontier
43:45
large
43:46
language
43:46
model
43:47
and
43:47
that
43:47
would
43:48
effectively
43:48
allow
43:49
you
43:49
to
43:49
execute
43:50
open
43:50
clause
43:50
sort
43:51
of,
43:51
you
43:51
know,
43:51
diff
43:51
different
43:52
agents.
43:52
Maybe
43:53
one
43:53
that's
43:53
always
43:53
listening,
43:54
one
43:54
that's
43:55
connected
43:55
to
43:55
your,
43:55
I
43:56
don't
43:56
know,
43:56
your
43:56
WhatsApp
43:57
API,
43:57
your,
43:57
your
43:58
SIG,
43:58
your
43:59
Slack
43:59
API
44:00
maybe
44:01
and
44:01
so
44:01
on
44:01
and
44:02
start
44:02
to
44:02
run
44:03
and
44:03
execute
44:04
tasks
44:04
autonomously
44:05
for
44:05
you.
44:05
I
44:06
think
44:06
I
44:06
saw
44:06
a
44:06
breakdown
44:06
of
44:07
it
44:07
the
44:07
other
44:07
day
44:07
and
44:07
actually
44:08
it's
44:08
cheapest
44:09
using
44:09
one
44:09
of
44:09
the
44:09
biggest
44:10
Chinese,
44:11
most
44:11
sophisticated
44:12
Chinese
44:13
models
44:13
that's
44:13
just
44:14
come
44:14
out
44:14
that'
44:14
fully
44:15
open
44:15
source
44:15
and
44:15
state
44:15
of
44:16
the
44:16
art.
44:16
It
44:17
would
44:17
cost
44:17
you
44:17
roughly
44:18
about
44:18
$2,000
44:20
a
44:20
year
44:21
to
44:21
run
44:21
it
44:21
as
44:21
a
44:22
personal
44:22
assistant.
44:23
So
44:23
capability
44:24
wise
44:24
it's
44:25
incredible.
44:26
Where
44:26
all
44:26
security
44:27
wise
44:27
it
44:28
isn't.
44:28
Yeah,
44:29
security,
44:29
it's
44:30
terrible.
44:30
And
44:30
that's
44:31
where
44:31
all
44:31
the
44:31
complaints
44:31
are
44:32
coming
44:32
from.
44:32
Now
44:33
if
44:33
we
44:33
go
44:33
back
44:33
to
44:33
education
44:34
and
44:34
schools.
44:35
Yeah,
44:35
it's
44:36
a
44:36
non
44:36
starter
44:36
in
44:37
its
44:37
current
44:37
implementation.
44:38
But
44:39
is
44:39
it
44:39
useful?
44:40
Is
44:40
it
44:40
an
44:40
interesting
44:41
concept?
44:41
Very
44:42
often,
44:42
you
44:42
know,
44:42
Mikey,
44:43
what
44:43
comes
44:43
out
44:43
of
44:43
these
44:43
things
44:44
are
44:44
slightly
44:44
flawed
44:45
implement.
44:46
Not
44:47
to
44:47
say
44:47
it's
44:47
not
44:47
impressive,
44:48
but
44:48
it's
44:48
all
44:48
at
44:48
like
44:49
a
44:49
proof
44:49
of
44:49
concept
44:50
level.
44:50
Just
44:51
like
44:51
write
44:52
it,
44:53
put
44:53
it
44:53
on
44:53
GitHub
44:54
or
44:54
something
44:54
and
44:54
send
44:55
it
44:55
out
44:55
into
44:55
the
44:55
world
44:55
and
44:56
see
44:56
what
44:56
happens.
44:56
Cybersecurity,
44:57
maybe
44:58
not
44:58
the
44:58
sexiest
44:59
topic.
44:59
I
44:59
actually
44:59
used
45:00
to
45:00
work
45:00
in
45:00
it
45:00
for
45:00
a
45:00
few
45:01
years,
45:01
but
45:02
you
45:02
know,
45:02
it's
45:03
not
45:03
going
45:03
to
45:03
be
45:03
at
45:03
the
45:03
forefront
45:04
of
45:04
people's
45:04
minds,
45:05
but
45:05
they're
45:05
just.
45:05
Going
45:06
to
45:06
be
45:06
excited
45:06
about
45:06
a
45:07
new
45:07
thing
45:07
rather
45:08
than
45:08
really
45:08
thinking
45:09
through
45:09
like,
45:09
do
45:10
you
45:10
want
45:10
to
45:10
give
45:10
it
45:10
access
45:11
to
45:11
your
45:11
phone,
45:11
to
45:11
your
45:12
documents?
45:13
No,
45:14
you
45:14
shouldn't.
45:15
Until,
45:17
Until.
45:17
And
45:18
which
45:18
brings
45:18
me
45:18
to
45:19
an
45:19
amazing
45:19
point.
45:19
Right.
45:20
When
45:20
we
45:20
talked
45:21
last
45:21
time,
45:21
you
45:21
talked
45:22
about
45:22
this
45:22
idea
45:22
of
45:23
AI
45:24
and
45:24
your
45:24
knowledge
45:25
of
45:25
large
45:26
language
45:26
models
45:27
and
45:28
effectively
45:29
how
45:30
there's
45:30
kind
45:30
of
45:30
a
45:30
gap,
45:31
obviously.
45:31
Right.
45:32
So
45:32
there's
45:32
like
45:32
run
45:32
of
45:33
the
45:33
mill
45:34
to
45:34
you.
45:34
As
45:35
in
45:35
like
45:35
you
45:35
just
45:35
get
45:36
it.
45:36
It's
45:37
like
45:37
sci
45:37
fi
45:37
to
45:37
that.
45:39
How
45:39
do
45:39
you
45:40
and
45:40
that.
45:40
I'm
45:40
just
45:40
using
45:41
your
45:41
words,
45:41
which
45:41
I
45:41
thought
45:42
was
45:42
a
45:42
great
45:42
explanation.
45:43
Right.
45:43
You,
45:43
you
45:43
understand
45:44
the
45:44
technology
45:44
really
45:45
well.
45:45
They,
45:46
they've
45:47
heard
45:47
about
45:47
it.
45:47
And
45:47
a
45:48
great
45:48
example
45:48
is
45:48
the
45:49
article
45:49
that
45:50
you
45:50
just
45:50
talked
45:50
about.
45:51
They
45:51
used
45:51
AI.
45:52
People
45:52
talk
45:53
about
45:53
AI
45:53
like
45:54
it's
45:54
water.
45:55
Like
45:55
as
45:55
in
45:55
it's
45:55
never
45:56
explained.
45:56
Right.
45:57
But
45:57
it's
45:58
like
45:58
it's
45:58
got
45:59
AI.
45:59
Is
45:59
that
46:00
1%,
46:00
2%?
46:01
How
46:01
do
46:01
we
46:01
calculate
46:02
that?
46:02
Like,
46:02
do
46:02
you
46:03
know
46:03
what
46:03
I
46:03
mean?
46:03
They're
46:03
very
46:04
different.
46:04
Different.
46:04
But
46:05
the
46:05
interesting
46:05
question
46:06
I
46:06
have
46:06
for
46:06
founders
46:07
listening
46:08
is
46:08
how
46:09
do
46:10
you
46:10
stop
46:11
yourself
46:11
from
46:12
over
46:12
engineering
46:13
the
46:13
product
46:14
when
46:14
you
46:14
are
46:15
at
46:15
like
46:16
mile
46:17
24
46:17
of
46:17
25
46:18
and
46:19
people
46:19
are
46:19
at,
46:20
you
46:20
know,
46:20
the
46:20
start
46:20
line?
46:21
It's
46:21
a
46:21
good
46:21
question.
46:22
I
46:22
think
46:22
it
46:23
comes
46:23
down
46:23
to
46:23
experience.
46:24
So
46:24
I
46:24
know
46:24
the
46:25
term
46:25
Vibe
46:25
coding
46:26
has
46:26
taken
46:26
off
46:26
and
46:26
it's
46:27
this
46:27
very
46:28
broad,
46:28
very
46:28
funny.
46:29
Broad
46:30
is
46:30
extremely
46:31
broad,
46:31
isn't
46:31
it?
46:32
I
46:32
think
46:32
again,
46:32
like,
46:32
like,
46:33
like
46:33
OpenClore
46:34
or
46:34
like
46:34
any
46:35
AI
46:35
framework,
46:36
once
46:36
it's
46:36
out
46:36
there,
46:36
it's
46:37
very
46:37
difficult
46:37
to
46:37
contain
46:38
the
46:38
term
46:38
and
46:38
it
46:38
sort
46:38
of
46:38
morphs.
46:39
But
46:39
the
46:39
term
46:39
Vibe
46:40
coding
46:40
was
46:40
created
46:41
by
46:41
Andrej
46:41
Karpathy,
46:42
who
46:42
is,
46:43
sorry,
46:43
I
46:43
hope
46:43
I
46:43
pronounced
46:44
that
46:44
Correctly,
46:44
he's
46:45
a
46:45
former
46:46
head
46:46
of
46:47
AI
46:47
at
46:47
Tesla
46:48
during
46:48
slightly
46:49
earlier
46:49
days
46:49
and
46:50
he
46:50
coined
46:51
the
46:51
term
46:51
Vibe
46:52
coding.
46:52
Now
46:52
he's
46:52
a
46:53
man
46:53
who
46:53
knows
46:53
what
46:53
he's
46:54
doing.
46:54
He
46:54
has
46:55
coded
46:55
manually
46:56
in
46:56
the
46:56
past.
46:56
He's
46:56
done
46:56
it
46:57
for
46:57
years
46:57
and
46:57
years
46:57
and
46:57
years.
46:58
So
46:58
for
46:58
him
46:58
it
46:59
was
46:59
just
46:59
kind
46:59
of
46:59
a
46:59
fun
47:00
experiment
47:00
to
47:00
sit
47:01
back
47:01
in
47:01
his
47:01
chair,
47:01
click
47:02
on.
47:02
I
47:03
think
47:03
he
47:03
was
47:03
using
47:03
a
47:03
microphone
47:04
effectively
47:05
to.
47:05
And
47:06
maybe
47:06
was
47:06
it
47:07
a
47:07
software
47:07
called
47:07
Super
47:07
Whisperer
47:08
or
47:08
something
47:08
for
47:09
Max
47:09
maybe.
47:10
But
47:10
effectively
47:10
he
47:11
would
47:11
speak
47:12
his
47:12
instructions
47:13
in
47:13
and
47:14
just
47:14
hit
47:14
enter
47:14
on
47:15
his
47:15
keyboard
47:15
and
47:16
let
47:16
it
47:16
execute.
47:16
And
47:17
for
47:17
him
47:17
it
47:17
was
47:17
just
47:17
a
47:17
bit
47:18
of
47:18
fun.
47:18
It
47:18
was,
47:19
let's
47:19
see
47:19
what
47:19
it
47:19
does,
47:20
what
47:20
it
47:20
gets
47:20
right,
47:21
what
47:21
it
47:21
gets
47:21
wrong.
47:22
Well,
47:22
this
47:22
is
47:22
actually
47:22
quite
47:23
impressive.
47:23
The
47:23
internal
47:24
reasoning
47:24
capability.
47:25
What
47:25
Vibe
47:26
coding
47:26
has
47:26
morphed
47:26
into
47:27
now
47:27
is
47:27
I
47:27
think
47:28
a
47:28
sort
47:28
of.
47:29
It's
47:29
like
47:29
this
47:29
battleground
47:30
between
47:30
engineers
47:31
and
47:31
non
47:31
engineers.
47:31
Engineers
47:32
where
47:32
non
47:32
engineers
47:32
say,
47:33
why
47:33
do
47:33
I
47:33
need
47:33
engineers?
47:34
I
47:34
can
47:34
just
47:34
build
47:34
this
47:35
myself.
47:35
And
47:36
then
47:36
they
47:36
get
47:36
to
47:36
a
47:36
particular
47:37
function
47:37
and
47:37
it's
47:38
horribly
47:38
insecure,
47:38
it
47:39
doesn't
47:39
work,
47:39
and
47:40
so
47:40
on.
47:40
And
47:40
there's
47:40
a
47:40
new
47:41
term
47:41
emerging
47:41
again
47:42
come
47:42
from
47:42
same
47:43
person,
47:43
Andrej
47:43
Karpathy.
47:44
And
47:44
it's
47:45
now
47:45
it's
47:45
called
47:45
agentic
47:46
engineering.
47:47
And
47:47
I
47:47
think
47:47
the
47:48
difference
47:48
between
47:48
Vibe
47:49
coding
47:49
and
47:49
agentic
47:50
engineering
47:50
is
47:51
respectfully
47:51
Vibe
47:52
coding.
47:52
Someone
47:52
who
47:52
doesn't
47:53
have
47:53
a
47:53
technical
47:53
background,
47:54
somebody
47:54
who
47:54
sits
47:55
on
47:55
Plit
47:55
Lovable
47:56
and
47:56
all
47:56
those
47:56
sorts
47:56
of
47:57
things,
47:57
those
47:57
sites
47:58
and
47:58
starts
47:59
providing
47:59
instructions.
48:00
But
48:00
when
48:00
something
48:00
goes
48:01
wrong,
48:01
they're
48:01
not
48:01
going
48:01
to
48:01
look
48:02
at
48:02
the
48:02
code.
48:02
They
48:03
might
48:03
just
48:03
ask
48:04
for
48:04
single
48:05
fixes.
48:05
And
48:05
the
48:05
difference
48:06
between
48:06
that
48:06
and
48:07
agentic
48:07
coding
48:07
is
48:08
somebody
48:08
who's
48:08
had
48:08
the
48:09
experience,
48:09
somebody
48:09
who's
48:10
gone
48:10
through
48:10
the
48:10
pain
48:10
of
48:11
it,
48:11
who
48:11
understands
48:12
what
48:12
best
48:12
practice
48:13
looks
48:13
like,
48:14
who
48:14
sets
48:15
up
48:15
a
48:16
framework
48:17
for
48:17
the
48:18
way
48:18
that
48:19
their
48:20
agents
48:21
should
48:21
go
48:21
out
48:22
and
48:22
execute
48:22
tasks.
48:23
How
48:23
do
48:23
they
48:23
update
48:24
documentation
48:25
afterwards?
48:25
How
48:25
do
48:26
they
48:26
do
48:26
so
48:26
in
48:26
a
48:27
way
48:27
that
48:27
one
48:27
agent
48:28
knows
48:28
what
48:28
another
48:28
agent
48:29
is
48:29
doing.
48:29
You
48:29
know,
48:29
does
48:30
one
48:30
agent
48:30
go
48:30
and
48:30
explore
48:31
something
48:31
another
48:31
agent
48:32
executes?
48:32
Did
48:33
you
48:33
split
48:33
them
48:33
out
48:33
that
48:34
way?
48:34
Does
48:35
one
48:35
agent
48:35
go
48:36
and
48:36
update
48:36
the
48:37
documentation
48:37
while
48:37
the
48:38
other
48:38
one
48:38
just
48:38
waits
48:38
to
48:39
read
48:39
the
48:39
updated
48:39
documentation
48:40
to
48:40
go
48:41
before
48:41
it
48:41
can
48:41
go
48:41
and
48:41
do
48:42
something
48:42
else?
48:42
You
48:43
know,
48:43
so
48:43
it.
48:44
If
48:44
there's
48:44
one
48:45
thing
48:45
I
48:46
can
48:46
give
48:46
to
48:46
your
48:46
listeners,
48:46
it
48:47
would
48:47
be
48:47
if
48:48
you
48:48
want
48:48
to
48:48
move
48:48
from
48:49
being
48:49
a
48:49
Vibe
48:49
coder
48:50
to
48:51
a.
48:51
An
48:51
agent
48:52
engineer,
48:54
then
48:54
start.
48:55
Are
48:56
trying
48:56
to
48:56
understand
48:57
a
48:58
systems
48:58
thinking
48:59
approach.
49:00
So
49:01
how
49:01
think
49:02
of
49:02
everything
49:02
like
49:02
a
49:03
box.
49:03
How
49:04
do
49:04
boxes
49:05
connect
49:05
to
49:05
other
49:05
boxes?
49:06
How
49:06
big
49:06
are
49:07
the
49:07
boxes?
49:07
So
49:07
how
49:08
large
49:08
or
49:08
small
49:08
Are
49:08
the
49:08
components,
49:09
how
49:09
much
49:09
are
49:09
they
49:10
responsible
49:10
for?
49:10
What
49:10
do
49:10
they
49:11
do?
49:11
What
49:11
boxes
49:12
are
49:12
contained
49:12
within
49:12
other
49:13
boxes?
49:13
So
49:13
what's
49:14
a
49:14
subset
49:14
or
49:15
of
49:16
what
49:16
else
49:16
a
49:17
function
49:17
of,
49:17
you
49:17
know,
49:18
what's
49:18
a
49:18
small
49:18
function
49:19
perhaps
49:19
of
49:19
one
49:20
larger
49:20
module
49:21
or
49:21
bit
49:21
of
49:21
software?
49:22
How
49:22
does
49:22
it
49:23
all
49:23
connect
49:23
together?
49:24
And
49:24
so
49:25
I
49:25
think
49:25
that's,
49:25
that's
49:26
a
49:26
big
49:26
difference,
49:26
you
49:27
know,
49:27
so
49:27
knowing
49:28
when
49:29
not
49:29
to
49:29
over
49:29
engineer
49:30
is
49:30
sometimes
49:31
when
49:31
you
49:31
see
49:32
something,
49:32
the
49:32
AI
49:33
doing
49:33
something
49:33
silly,
49:34
you
49:34
think
49:35
that's
49:35
just
49:35
a
49:35
really
49:35
simple
49:36
bug,
49:36
I
49:36
could
49:36
just
49:36
go
49:36
into
49:37
the
49:37
code
49:37
and
49:37
change
49:37
that.
49:38
Or
49:38
I
49:38
just
49:38
need
49:38
the
49:39
AI
49:39
to
49:39
focus
49:39
on
49:39
this
49:40
one
49:40
little
49:40
thing
49:41
or
49:41
I
49:41
just
49:41
need
49:42
to
49:42
set
49:42
up
49:42
tests
49:42
for
49:43
that
49:43
little
49:44
workflow
49:44
so
49:44
that
49:45
that
49:45
doesn't
49:45
happen
49:45
again
49:46
when
49:46
ends
49:47
Experience.
49:48
Yeah,
49:48
yeah,
49:49
so
49:49
it's
49:50
definitely
49:50
experience
49:51
and
49:51
it's
49:51
that
49:51
kind
49:51
of.
49:52
I
49:52
always
49:52
talk
49:52
about
49:53
if,
49:53
if
49:53
this,
49:53
then
49:54
that
49:54
in
49:54
the
49:54
simplest
49:55
form,
49:55
right.
49:55
With
49:56
systems
49:56
thinking,
49:57
if
49:57
I
49:58
do
49:58
this,
49:58
then
49:58
that
49:59
just,
49:59
you
49:59
know,
50:00
back
50:00
and
50:00
forth.
50:00
For
50:01
the
50:01
record,
50:02
Attish
50:02
did
50:03
nod
50:03
his
50:03
head
50:03
at
50:04
that.
50:05
I
50:06
said
50:06
to
50:06
a
50:06
guy
50:06
who
50:06
built
50:07
LLMs
50:08
system
50:08
thinking
50:09
and
50:09
logic.
50:09
Keep
50:10
it
50:10
simple.
50:10
Just
50:11
to
50:11
wrap
50:11
things
50:11
up
50:12
here,
50:12
buddy,
50:12
and
50:12
I'd
50:13
love
50:13
to
50:13
have
50:13
you
50:13
back
50:13
again
50:14
as
50:14
things
50:14
develop
50:15
with
50:15
pedagog
50:16
dog.
50:16
Yeah.
50:17
And
50:17
the,
50:18
the
50:18
industry
50:19
and
50:19
AI
50:20
in
50:20
general.
50:20
In
50:21
terms
50:21
of
50:22
your
50:22
sort
50:23
of
50:23
advice,
50:24
given
50:24
that
50:24
you
50:25
are
50:25
a
50:25
reservist
50:26
in
50:26
the
50:26
British
50:27
army,
50:27
you
50:27
have
50:27
such
50:28
an
50:28
interesting
50:28
background,
50:30
what
50:30
one
50:31
thing
50:31
would
50:31
you
50:32
say
50:32
makes
50:33
you
50:33
a
50:33
better
50:34
founder?
50:35
That's
50:36
not
50:36
a
50:36
technical
50:37
skill.
50:37
Okay.
50:38
So
50:39
I
50:40
think
50:41
this
50:41
would,
50:41
would
50:42
be
50:43
working
50:43
out
50:43
where
50:44
to
50:44
prioritize
50:44
your
50:45
time.
50:45
I
50:45
think
50:45
AI
50:46
has
50:47
made
50:47
things
50:48
even
50:49
easier,
50:50
as
50:50
if
50:50
we
50:50
didn't
50:51
have
50:51
it
50:51
already
50:51
in
50:52
the
50:52
world,
50:52
convenience.
50:53
So
50:53
I
50:53
think
50:54
it's
50:54
made
50:54
convenience
50:55
even
50:55
easier.
50:56
The
50:56
idea
50:57
that
50:57
you
50:57
can
50:57
go
50:58
from
50:58
thinking
50:58
about
50:58
something
50:59
in
50:59
natural
50:59
language
50:59
or
51:00
specifically
51:01
English,
51:01
that's
51:01
just
51:02
where
51:02
the
51:02
most
51:02
prevalent
51:03
AIs
51:03
live
51:03
at
51:03
the
51:04
moment,
51:04
with
51:04
the
51:04
English
51:04
language.
51:05
But
51:05
you
51:05
can
51:05
go
51:05
from
51:06
thinking
51:06
about
51:06
something
51:07
in
51:07
English
51:07
and
51:08
natural
51:08
language
51:08
to
51:09
having
51:09
it
51:09
in
51:09
front
51:09
of
51:10
you
51:10
in
51:10
five
51:10
minutes.
51:11
And
51:11
that's
51:11
unprecedented
51:12
for
51:13
as
51:13
long
51:13
as
51:13
the
51:13
software
51:14
engineering
51:14
industry
51:15
has
51:15
existed
51:15
and
51:16
so
51:16
on.
51:17
But
51:18
I
51:19
think
51:19
sometimes
51:19
it's
51:19
very
51:20
easy
51:20
to
51:20
get
51:20
caught
51:21
up
51:21
in
51:21
that
51:21
deliver,
51:22
deliver,
51:22
deliver,
51:22
deliver
51:23
mentality.
51:24
And
51:24
Certainly
51:24
that's
51:24
what
51:25
VCs
51:25
might
51:26
want
51:26
from
51:26
you
51:26
and
51:27
understandably
51:27
so
51:28
they've
51:28
made
51:28
an
51:28
investment.
51:29
I
51:29
think
51:30
sometimes
51:30
if
51:30
you
51:31
spend
51:31
time
51:31
in
51:32
processes
51:32
outside
51:33
of
51:33
software.
51:34
So
51:34
army
51:35
is
51:35
a
51:35
great
51:35
example.
51:36
If
51:36
you're
51:36
outside
51:37
somewhere,
51:37
somewhere
51:37
and
51:37
it's
51:38
raining,
51:39
you
51:39
can't
51:39
vibe
51:39
code
51:40
your
51:40
way
51:40
out
51:40
of
51:40
that.
51:40
You
51:41
just
51:41
need
51:41
to
51:43
problem
51:43
Solve
51:43
it.
51:44
You
51:44
need
51:44
to
51:44
put
51:44
up
51:44
your
51:44
tent
51:44
with
51:45
limited
51:45
light.
51:45
If
51:46
it's
51:46
dark,
51:46
you
51:47
can't
51:47
use
51:47
white
51:47
light
51:47
because
51:48
that
51:48
might
51:48
be
51:48
seen
51:49
by
51:49
an
51:49
opponent,
51:50
you
51:50
know,
51:50
an
51:50
enemy
51:51
in
51:51
the
51:51
field.
51:51
So
51:51
you've
51:51
got
51:51
to
51:51
use
51:52
red
51:52
light,
51:52
dim,
51:53
a
51:53
small
51:53
dim
51:53
red
51:54
light
51:54
to
51:54
put
51:54
up
51:55
a
51:55
tent
51:55
when
51:55
you're
51:55
already
51:56
soaked,
51:56
you're
51:57
really
51:57
tired,
51:57
you're
51:58
hurting,
51:58
but
51:58
you've
51:59
just
51:59
got
51:59
to
51:59
take
51:59
your
51:59
time
52:00
and
52:00
build
52:00
it
52:00
out
52:01
and
52:01
do
52:01
it
52:01
properly.
52:02
And
52:02
so
52:02
I
52:02
think
52:03
sometimes
52:03
spending
52:04
time
52:05
with
52:06
activities
52:07
and
52:07
exposing
52:08
yourself
52:08
to
52:09
activities
52:10
outside
52:10
of
52:10
and
52:11
away
52:11
from
52:11
the
52:11
digital
52:12
world,
52:12
where
52:12
the
52:13
expectation
52:13
is
52:14
just
52:14
everything
52:14
now,
52:15
can
52:16
really
52:16
help
52:16
you
52:17
build.
52:18
Yeah,
52:18
it
52:18
can
52:18
really
52:19
help
52:19
you
52:19
build
52:20
a
52:20
more
52:21
deliberate
52:22
mindset.
52:23
So
52:23
a
52:24
mindset
52:24
in
52:24
which
52:25
your
52:25
thinking
52:25
is
52:26
more
52:26
deliberate,
52:27
in
52:28
which
52:28
you
52:29
have
52:29
a
52:29
certain
52:29
amount
52:30
of
52:30
time
52:30
and
52:31
energy
52:31
to
52:32
focus
52:32
on
52:33
doing
52:34
a
52:34
thing
52:34
or
52:35
a
52:35
set
52:35
of
52:36
things.
52:36
And
52:37
my
52:37
advice
52:37
would
52:38
be
52:38
to
52:38
try
52:38
and
52:38
stick
52:38
with
52:39
something.
52:39
So
52:39
I
52:39
know
52:40
Naval
52:40
Ravikant,
52:41
who's
52:41
the
52:42
famous
52:43
angel
52:43
investor
52:44
in
52:44
the
52:44
States.
52:45
I
52:45
remember
52:46
him
52:46
saying
52:46
on
52:46
a
52:47
podcast,
52:47
I
52:47
think
52:47
it
52:48
was
52:48
around
52:48
Covid
52:48
time
52:48
when
52:49
I
52:49
was
52:49
going
52:49
for
52:49
a
52:49
walk,
52:50
he
52:50
said,
52:50
as
52:51
I
52:51
get
52:51
older,
52:51
the
52:51
only
52:52
thing
52:52
I
52:52
invest
52:52
in
52:53
is
52:53
compounding
52:55
things,
52:55
and
52:56
particularly
52:56
relationships.
52:58
So
52:58
that
52:59
which
52:59
is,
53:00
or
53:00
that
53:01
which
53:01
are
53:01
long
53:01
term
53:02
and
53:02
that
53:02
will
53:03
build
53:03
upon
53:04
themselves
53:04
recursively
53:05
over
53:06
time
53:06
and
53:06
improve,
53:07
get
53:07
stronger,
53:08
get
53:08
better,
53:08
bring
53:09
me
53:09
more
53:09
satisfaction,
53:10
increase
53:10
my
53:11
skill
53:11
level.
53:12
So
53:12
I
53:12
would
53:12
say
53:13
try
53:13
to
53:13
spend
53:13
time
53:14
with
53:15
a
53:15
few
53:15
set
53:16
things
53:16
that
53:16
you
53:17
can
53:17
progress
53:18
over
53:18
time,
53:19
that
53:19
you
53:19
can
53:19
show
53:19
your
53:20
proficiency
53:20
level
53:20
at,
53:21
and
53:21
so
53:21
on.
53:22
And
53:22
how
53:22
does
53:22
it
53:22
all
53:23
translate
53:23
back
53:23
to
53:24
software
53:24
engineering?
53:25
Well,
53:25
again,
53:26
I
53:26
think
53:27
everything
53:27
is
53:27
so
53:27
ubiquitous
53:28
and
53:28
on
53:28
demand
53:29
now
53:30
in
53:30
the
53:30
digital
53:31
space
53:32
that
53:32
it
53:32
actually
53:32
forces
53:33
you
53:33
to
53:33
stop
53:34
and
53:34
think,
53:35
am
53:35
I
53:36
making
53:36
a
53:36
good
53:36
decision?
53:37
Am
53:38
I
53:38
making
53:38
a
53:38
good
53:38
decision
53:39
in
53:39
context
53:40
of
53:41
whatever
53:42
it
53:42
is
53:42
from
53:43
a
53:44
human
53:44
perspective,
53:45
from
53:45
a
53:45
customer
53:46
perspective,
53:46
from
53:47
a
53:47
technical
53:47
design
53:48
perspective,
53:49
from
53:49
a.
53:49
Is
53:50
this
53:50
worth
53:50
my
53:51
time?
53:51
Should
53:51
I
53:52
be
53:52
doing
53:52
it?
53:52
Should
53:52
it
53:53
be
53:53
delegated?
53:53
Should
53:53
it
53:54
not
53:54
be
53:54
done
53:54
at
53:54
all?
53:55
Should
53:55
it
53:55
be
53:55
deprioritized?
53:56
And.
53:57
And
53:58
I
53:58
think
53:58
that
53:59
really
53:59
helps
53:59
with
54:00
what
54:00
I
54:00
call
54:00
holistic
54:01
thinking
54:01
at
54:02
least.
54:03
So,
54:03
yeah,
54:04
that
54:04
would
54:04
be.
54:04
My
54:05
advice
54:06
is
54:07
find
54:09
areas
54:09
in
54:10
which.
54:10
So
54:11
two
54:11
things
54:11
really.
54:12
One
54:13
areas
54:13
away
54:13
from
54:14
the
54:14
digital
54:14
space
54:15
where
54:15
the
54:15
laws
54:15
of
54:15
nature
54:16
still
54:16
apply.
54:16
Because
54:16
it
54:17
touch
54:17
grass,
54:17
as
54:18
they
54:18
say
54:18
these
54:18
days.
54:18
Touch
54:19
grass,
54:19
exactly.
54:20
Deliberate
54:20
thinking
54:21
mindset.
54:22
And
54:22
then
54:22
when
54:22
you
54:22
are
54:23
deliberately
54:23
thinking,
54:24
think
54:24
long
54:24
term
54:24
and
54:25
big
54:25
picture
54:26
and
54:26
vision
54:26
and
54:26
that
54:27
sort
54:27
of
54:27
thing.
54:27
So
54:27
think
54:28
about
54:29
where
54:29
you're
54:30
compounding
54:30
relationships
54:32
or
54:33
where
54:33
the
54:33
compounding
54:34
factors
54:34
are
54:35
like
54:35
just,
54:35
just
54:35
try
54:35
and
54:36
think.
54:36
If
54:36
I
54:36
Do
54:36
this
54:37
now.
54:38
Where
54:38
can
54:38
I
54:38
be
54:38
in
54:39
three
54:39
or
54:39
five
54:39
years?
54:40
From
54:40
there,
54:40
those
54:40
two
54:41
things
54:41
really
54:41
feed
54:41
into
54:42
each
54:42
other.
54:42
Deliberate
54:42
thinking
54:43
and
54:43
compounding.
54:44
I
54:45
think
54:45
that's
54:45
excellent
54:46
advice.
54:46
I
54:46
think
54:47
I,
54:47
you
54:47
know,
54:47
if
54:47
I
54:47
was
54:47
to
54:48
summarize
54:48
what
54:48
you
54:48
said
54:49
in
54:49
like,
54:49
like
54:50
five
54:50
words,
54:51
four.
54:51
I'd
54:52
say
54:52
think
54:53
slow.
54:53
Act
54:53
fast.
54:54
As
54:54
in
54:54
think
54:55
slow
54:55
first.
54:57
You
54:58
can
54:58
act
54:58
fast
54:59
with
54:59
technology
55:00
now
55:00
anyway.
55:01
But
55:01
if
55:01
you
55:02
start
55:03
in
55:03
the
55:03
wrong
55:03
direction
55:04
to
55:04
begin
55:04
with
55:04
and
55:05
you
55:05
don't
55:05
really
55:05
think
55:06
it
55:06
through,
55:06
and
55:07
I
55:07
think
55:07
that
55:07
is
55:07
the
55:08
biggest
55:08
problem
55:08
nowadays
55:09
is
55:09
like,
55:10
oh,
55:10
I've
55:10
got
55:10
this
55:11
idea
55:12
and
55:12
then,
55:14
oh,
55:15
I'm
55:15
so
55:15
far
55:15
down
55:16
the
55:16
road
55:16
and
55:17
I've
55:17
now
55:18
got
55:18
pulled
55:18
in
55:18
a
55:18
different
55:19
direction.
55:19
Another
55:19
direction,
55:20
another
55:20
direction.
55:21
And
55:21
AI
55:22
as
55:22
I
55:23
understand
55:23
it
55:24
from
55:24
a
55:24
large
55:24
language
55:25
model
55:25
point
55:25
of
55:25
view
55:25
works
55:26
on
55:26
the
55:26
basis
55:26
of
55:26
probabilities.
55:27
Yes.
55:28
And
55:28
it
55:28
is.
55:28
So
55:29
it's
55:29
never
55:29
going
55:29
to
55:30
be
55:30
a
55:30
hundred
55:30
percent
55:31
right.
55:32
It's
55:32
never
55:32
going
55:32
to
55:33
be
55:33
a
55:33
hundred
55:33
percent
55:33
consistent.
55:35
It
55:35
will
55:35
do
55:35
its
55:36
best.
55:36
You
55:37
have
55:37
to
55:37
really
55:38
I,
55:38
the
55:38
biggest
55:39
worry
55:39
I
55:39
have,
55:39
mate,
55:40
and
55:40
we
55:40
can
55:40
talk
55:40
about
55:40
this
55:41
next
55:41
time,
55:41
but
55:42
is
55:42
that
55:42
people
55:43
are,
55:43
are.
55:44
What's
55:44
the
55:44
word
55:44
I'm
55:44
on
55:44
working,
55:44
looking
55:45
for.
55:45
They're
55:46
outsourcing
55:47
thinking
55:49
through
55:50
technology,
55:51
as
55:52
in
55:52
big
55:53
picture
55:53
thinking.
55:54
And
55:55
so
55:55
like
55:55
if
55:55
we
55:56
get
55:56
to
55:56
a
55:56
stage
55:56
where
55:57
it's
55:57
like,
55:57
you
55:57
know,
55:57
how
55:57
do
55:58
I
55:58
make
55:58
a
55:58
sandwich?
55:59
You
55:59
know,
56:00
like,
56:00
come
56:00
on,
56:01
use,
56:01
you
56:01
know,
56:02
use
56:03
what
56:03
we've
56:03
all
56:04
got.
56:05
Because
56:05
you
56:05
know,
56:06
getting
56:06
hands
56:06
on
56:07
with
56:07
things
56:07
I
56:07
think
56:07
is
56:07
a
56:08
great.
56:08
And
56:08
to
56:09
your
56:09
point
56:09
about
56:09
the
56:10
tent
56:10
is.
56:10
And
56:11
facing
56:11
challenges
56:12
and
56:12
problem
56:12
solving,
56:13
I
56:13
highly
56:14
recommend
56:14
that
56:14
for
56:14
everybody
56:15
as
56:15
well.
56:15
And
56:15
don't
56:16
get
56:16
too
56:16
panicky.
56:17
Is
56:18
like
56:18
things
56:18
tend
56:19
to
56:19
be
56:19
overestimated
56:20
about
56:20
how
56:20
quick
56:21
they'll
56:21
change.
56:22
Yes.
56:24
It's
56:24
a
56:24
marathon,
56:25
not
56:25
a
56:25
sprint
56:26
for
56:26
sure.
56:26
Even
56:27
if
56:27
the
56:27
marathon
56:27
is
56:28
only
56:29
three
56:29
years
56:30
long
56:30
in
56:30
this
56:31
crazy
56:31
age
56:31
of
56:31
growing
56:32
companies
56:33
and
56:33
then
56:33
getting
56:33
them
56:34
acquired,
56:34
it's
56:34
still
56:34
a
56:35
marathon.
56:35
It's
56:36
not
56:36
like,
56:36
like
56:36
you'll
56:36
wake
56:37
up
56:37
literally
56:38
tomorrow
56:38
and
56:38
things
56:39
will
56:39
be
56:39
different.
56:39
And
56:39
the
56:40
last
56:40
two
56:40
things
56:40
I,
56:40
I'd
56:41
add
56:41
there.
56:41
So
56:41
I,
56:41
yeah,
56:42
I,
56:42
I
56:42
like
56:42
that
56:42
thing.
56:43
Slow,
56:43
act
56:43
fast.
56:44
In
56:44
fact,
56:44
if
56:45
you've
56:45
not
56:45
read
56:45
Daniel
56:45
Kahneman's
56:46
book
56:46
Thinking,
56:47
what
56:48
was
56:48
it?
56:48
Thinking
56:48
fast
56:48
and
56:49
slow.
56:50
Thinking
56:50
fast
56:50
and
56:50
slow.
56:51
Yeah.
56:51
Thinking
56:51
fast
56:51
and
56:52
slow.
56:52
Yeah,
56:52
I
56:52
think
56:52
he
56:52
passed
56:53
away
56:53
recently.
56:53
But
56:54
yeah,
56:54
excellent
56:55
book
56:55
and,
56:55
and
56:55
lauded
56:56
for
56:56
a
56:56
reason.
56:56
U.S.
56:57
navy
56:57
SEALs,
56:58
I
56:58
think
56:58
have
56:58
a
56:59
phrase
56:59
which
56:59
is
56:59
slow
57:00
is
57:00
smooth,
57:01
smooth
57:01
is
57:01
fast.
57:01
So,
57:02
so
57:03
deliberate
57:03
thinking.
57:04
And
57:04
that
57:04
will
57:04
build
57:05
up
57:05
over
57:05
time.
57:06
And
57:06
then
57:06
the
57:06
very
57:06
last
57:07
thing
57:07
I'd
57:07
say
57:07
in
57:07
terms
57:07
of
57:08
large
57:08
language
57:08
models
57:09
is
57:10
people
57:10
talk
57:10
about,
57:11
oh,
57:11
you've
57:11
got
57:11
all
57:11
this
57:12
knowledge
57:12
at
57:12
your
57:12
fingertips.
57:13
I
57:13
mean,
57:13
look,
57:14
if
57:14
you
57:14
think
57:14
of
57:14
problems
57:14
generally
57:15
there
57:15
are
57:15
known
57:15
knowns,
57:16
there
57:16
are
57:16
known
57:17
unknowns,
57:18
both
57:18
of
57:18
which
57:19
known
57:19
knowns
57:19
large
57:20
language
57:20
models
57:20
excel
57:21
at
57:21
because
57:21
they're
57:22
pre
57:22
trained
57:22
on
57:22
a
57:22
lot
57:22
of
57:23
data,
57:23
known
57:24
unknowns,
57:24
they
57:24
can
57:25
help
57:25
you
57:25
spot
57:25
gaps
57:25
and
57:26
logical
57:26
flaws.
57:27
And
57:27
unknown
57:28
unknowns.
57:29
How
57:29
do
57:29
you
57:29
even
57:29
know
57:29
you're
57:30
looking
57:30
for
57:30
them?
57:30
You
57:30
don't.
57:31
And
57:31
they
57:31
don't
57:31
exist
57:32
in
57:32
training
57:32
data.
57:33
So
57:34
will
57:35
they
57:35
ever
57:35
be
57:35
good
57:35
at
57:36
unknown
57:36
unknowns?
57:37
Who
57:37
knows?
57:38
Maybe,
57:38
potentially.
57:39
Maybe
57:39
it's
57:39
a
57:39
different
57:39
type
57:40
of
57:40
AI
57:40
that
57:40
will
57:40
do
57:41
that.
57:41
But
57:41
right
57:41
now
57:42
the
57:42
human
57:42
brain
57:43
is
57:43
the
57:44
unbeaten
57:44
champ
57:45
at
57:45
doing
57:45
that.
57:46
It
57:46
is
57:46
incredibly
57:47
dynamic,
57:47
incredibly
57:48
efficient
57:49
and
57:49
so
57:50
there
57:50
is
57:50
always
57:51
room
57:51
and
57:51
potential
57:52
for
57:52
human
57:52
creativity,
57:53
even
57:53
as
57:53
AI
57:54
continues
57:54
to
57:54
explore.
57:56
I
57:56
love
57:56
that
57:56
because
57:57
fundamentally
57:58
the
57:58
creativity
57:59
that
57:59
we
57:59
see
57:59
coming
57:59
out
58:00
of
58:00
generative
58:00
AI
58:01
is
58:01
based
58:01
on
58:03
training
58:03
data.
58:04
So
58:04
yes,
58:05
exactly.
58:06
It's
58:06
not
58:07
actually
58:07
novel
58:08
thinking,
58:08
don't
58:08
worry
58:09
script
58:09
writers
58:10
and
58:10
set
58:11
designers
58:11
just
58:11
yet.
58:12
But
58:12
yes,
58:13
I
58:13
thank
58:14
you
58:14
so
58:14
much
58:14
for
58:14
your
58:15
time
58:15
mate.
58:15
I
58:15
will
58:16
post
58:17
all
58:17
of
58:17
the
58:17
information
58:18
down
58:18
below
58:19
to
58:19
find
58:19
out
58:19
about
58:19
pedagogical
58:21
and
58:21
also
58:22
pop
58:22
your
58:22
details
58:23
down
58:23
there
58:23
as
58:23
well.
58:24
And
58:25
I
58:25
hope
58:25
to
58:25
have
58:25
you
58:25
back
58:26
again
58:26
in
58:26
the
58:27
future
58:27
just
58:27
to
58:28
see
58:28
how
58:28
the
58:29
world
58:29
of
58:29
education
58:29
is
58:30
developing
58:31
with
58:31
the
58:31
guy
58:32
who's
58:32
trying
58:32
to
58:33
set
58:33
the
58:33
gold
58:34
standard.
58:35
So
58:35
thank
58:35
you
58:35
so
58:35
much
58:35
for
58:35
your
58:36
time,
58:36
I
58:36
really
58:36
do
58:36
appreciate
58:37
it.