Latest / AI for Founders with Ryan Estes / Synthetic Relationships, FTW

Transcript

Synthetic Relationships, FTW · AI for Founders with Ryan Estes

0:00
What if the characters your kids grow up loving grow up with
0:03
them? What if Mario already knows your
0:05
kids name, their mood, and exactly what to say to keep them
0:08
coming back? And he's doing it across a
0:10
billion accounts simultaneously. I'm talking about Super Mario
0:12
Brothers. So here's what that means.
0:15
AI is about to replace your influencer budget and do it
0:17
better. Not from ads and not from
0:19
influencers. From the characters sliding into
0:22
your DMS like an old friend who just happens to know exactly
0:24
what game you've been neglecting.
0:26
And now these characters, like Mario, have personality and
0:30
conversations. And here's where it gets
0:32
interesting. This not just gaming or
0:34
entertainment. This is the distribution model
0:36
of the future. A branded character agent that
0:39
owns your DMS, warms your leads, and never sleeps.
0:42
Imagine your SAS mascot texting your churned users.
0:45
Imagine your e-commerce brands character knowing someone's vibe
0:48
before your sales team does and the business model revenue
0:52
share. I mean, what would you be
0:53
willing to pay if Mario can literally sell your customers?
0:56
So if AI can build a more consistent, more available, more
0:59
patient relationship with your customer than any human on your
1:02
team, should we start thinking about what characters most
1:05
embody our brands? And just pass the ball to them.
1:08
Welcome to AI for. Founders I'm Ryan Estes, and
1:10
that was Rebecca Lau, Co founder and CEO of Saga AI Labs.
1:15
In this interview, you're going to learn how synthetic character
1:17
agents are reshaping user acquisition.
1:20
This one's for founders in gaming, consumer brands,
1:23
entertainment, and anyone building a product that needs a
1:25
distribution edge. If you want the full breakdown
1:28
framework, founders tools, and the playbooks behind the
1:30
conversations like this one, sign up for AI for founders.co,
1:34
where we turn bleeding edge AI into moves you can actually make
1:37
Monday morning. It's free, it's founder 1st, and
1:40
honestly, your competitors probably haven't found it yet.
1:43
Let's keep it that way. If this episode made you think
1:46
differently about distribution, growth or your next product
1:48
decision, drop us a five star review on the podcast.
1:51
We read every single review and it genuinely helps us bring more
1:55
founders like Rebecca into the room with you.
1:58
And if you're building an audience on rented land, you're
2:01
1 algorithm change away from starting over.
2:04
inboxalchemy@inboxalchemy.co is the number one distribution
2:08
motion for founders who are done playing with someone else's
2:11
rules. We're talking 25,000 targeted
2:14
newsletter subscribers, your actual ICP delivered in your
2:18
first six months, weekly newsletters crafted for
2:21
founders, partnership opportunities that put your
2:23
brand in front of the right rooms, and a personal brand that
2:26
compounds like equity. You don't just reach your
2:29
audience, you own them for life. Stop renting attention.
2:33
Start building an asset. Go to inboxalchemy.co and let's
2:37
grow something that's actually yours.
2:39
That's Inbox alchemy.co. Rebecca, thank you so much for
2:49
coming on the show. It's great to meet you.
2:51
Great to meet you too, Ryan. Thanks so much for having me.
2:53
And we've been in the books with this podcast for a few months
2:57
now, so it's finally happening and I'm really excited.
3:00
It's happening. I, I, I'm excited too, you know,
3:04
and particularly because when I look at your website, which is
3:06
saga dot dot XYZ, the company's called Saga.
3:11
It's full of mystique and entry, you know, it's like, OK, there's
3:17
entertainment, there's gaming, there's AI, there's all these
3:20
different components. Fill us in on what Saga AI is.
3:25
Yeah. So Saga AI Labs specializes in
3:28
synthetic relationships, and in particular, we develop what's
3:34
known as character AI agents. So these are AI agents that are
3:38
based off of characters. They can come from gaming, from
3:42
film, TV, general entertainment, from sports brands branching
3:48
into consumer brands as well. And these characters are all
3:53
trained on the lore of that particular character.
3:56
So what is the background, what is the content around that
4:00
character already? What is the history of it, what
4:02
is the personality of that character, etcetera.
4:05
And these agents can be leveraged for several purposes
4:10
and so they can range from user acquisition.
4:12
That's a very big one, especially for the gaming
4:15
industry. So finding new users, retaining
4:17
users, reactivating users that have lapsed all the way to
4:22
community and fan engagement. So you already have a very
4:26
vibrant community, you just need to have a compelling set of
4:30
people and or agents to engage with that community.
4:34
And if you are already a fan of something, then talking to a
4:38
highly recognizable and beloved character from that particular
4:42
brand or IP, there's nothing like it.
4:45
So that is what we do. And the reason why we emphasize
4:48
that synthetic relationship is our goal is to make these
4:52
characters as lifelike as possible.
4:54
It should not feel like you were talking to an agent.
4:57
If at any point you know that you're talking to a chatbot, we
5:00
will have failed. So it needs to be a very natural
5:03
conversation. And the agents, you know, they
5:06
don't become negative. They don't ever attack the
5:10
users. So there are, you know, strict
5:12
guardrails that we put up, especially around brand
5:14
guidelines. But having said that, they do
5:16
have very rich and nuanced personalities.
5:18
A lot of that is owing to the original creative teams behind
5:21
these characters, and it usually leads to pretty satisfying and
5:24
compelling conversations. Now, once you're able to develop
5:28
that kind of relationship with a user, then you can do a lot with
5:31
the user, but it all starts with that relationship.
5:34
Synthetic relationships, Yes. Amazing.
5:38
So it would this be like if I'm like, streaming Nintendo?
5:43
Yeah. And I'm like, you know what?
5:44
I'm going to, I'm going to save a couple bucks this month,
5:46
cancel my subscription. Then a little head would pop up
5:49
and be like, oh Ryan, why you want to break up with a Mario?
5:54
That's cute. It's very, very cute.
5:56
And yes, that is definitely one use case.
6:01
It happens. It happens quite a lot,
6:03
especially if the characters are already in DMS with the users.
6:09
So we live across primarily all major social channels.
6:13
So instead, TikTok, Reddit X Meta, WhatsApp, Discord,
6:20
etcetera. And so there are conversations
6:24
that take place between these agents and users in public
6:27
settings. But really the richest
6:29
conversations happen in DM. And so if you already have that
6:33
DM channel open with the character, then yes, they will
6:36
be pinging to say, hey, you know, or you're breaking up with
6:39
me. But we're all so careful about
6:42
that because at the end of the day, this is meant to not feel
6:46
like traditional advertising or performance marketing.
6:48
We just never want to feel annoying or spammy.
6:52
So characters will, you know, reinitiate conversations.
6:57
They obviously will will try to re establish relationships with
7:02
users that have gone a little cold.
7:04
But at a certain point we do know not to push, to give a time
7:08
to have a reason to talk to you, etcetera.
7:11
But yes, you have the right idea.
7:12
Interesting, so the the communication happens in the DMS
7:16
of like an Instagram account? It does, yeah.
7:19
Go ahead, Go ahead. Yeah, yeah, the server would be
7:21
like what would be a good example of a character.
7:25
So 1 character that is live already is from Trivia Crack.
7:30
Trivia Crack is the number one trivia game in the world, has
7:33
about 800 million downloads throughout its life, and their
7:37
main mascot is Willie. So Willie is a spinny multi
7:41
colour trivia wheel and Willie lives across Insta and X
7:46
currently. And yes, he is available in the
7:50
DMS, but that's table stakes. The real magic of our agents is
7:54
in the ability to find new users.
7:56
So user acquisition is really what most studios are looking
8:01
for. And what that means is that the
8:02
agents are crawling across social media to identify users
8:08
that could potentially be interested in this particular
8:11
game. And that doesn't mean that the
8:14
users are mentioning the game or even saying the words trivia or
8:18
crack or whatever. It's those are two easy
8:20
giveaways. Usually what Willie is doing is
8:23
he is commenting on trending topics.
8:27
He's identifying users that are within the core demographics we
8:30
know for the game and engaging them so going into their replies
8:34
or common threads. And there Willie is very
8:37
successful. So we do get 90% engagement on
8:41
our commenting activity. So anytime that Willie is out
8:46
there, 90% of the time he gets peek likes on that thread that
8:51
he's commented on and people continue to engage with him.
8:54
So that speaks to the strength of the agents.
8:58
The next step is to really solidify conversion numbers and
9:03
yeah, then then the agents are are off doing their thing.
9:06
Cool. So how much consideration goes
9:08
into like, you know, comments? So if he's commenting, he's
9:12
chatting. There's certain things that are
9:14
going to like spawn engagement and then there's certain things
9:17
that will have to acquiesce to Willy's personality.
9:21
How do you kind of balance that? Yeah, so Willie if OK, so there
9:27
are brand guidelines around Willie.
9:29
So he can't talk about religion or politics.
9:32
He can't curse. So if a user tries to quote, UN
9:36
quote, break Willie or engages Willie in, you know, one of
9:39
those things that's going to violate brand guidelines, then
9:41
Willie will just in a very charming way, say, oh, you know,
9:43
I don't know anything about that or I can't really talk about
9:45
that. But it really the goal of the
9:51
agent is to make sure that they still hew to their personality.
9:55
But personalities are nuanced. So Willie, for instance, if a
10:00
user is a little more extroverted, then Willie goes
10:04
all out. You know, he is his usual perky
10:06
excited self all the time. If a user is engaged but maybe
10:10
doesn't have as many words, then Willie will match that energy
10:14
accordingly. So a little quieter, but the
10:17
personality remains the same. So it's, it's actually that part
10:23
of it is not as difficult as we anticipated is the, the agent
10:28
hewing to, to the personality. I think what the training model
10:34
will continue to to improve upon this is our area of focus is how
10:38
do we make sure that the agent continues to be compelling as a
10:43
conversationalist. So not just, you know, showing
10:46
up as themselves, which is, you know, very cute and fun, but how
10:49
do they lead the user to become a bigger part of the community
10:54
to find other people in the community to just know that IP
10:58
and that brand better etcetera. So that piece of it is is really
11:04
the the next big step. That's interesting.
11:07
So would it be like talking to Will and he's like, hey, Ryan
11:10
Nuggets play on Thursday, There's still some tickets.
11:12
Should we get some? Yeah, yeah, yeah, yeah, yeah,
11:16
absolutely. Quick side note for founders who
11:19
are using AI to write content, you've probably noticed this.
11:22
It's incredibly fast, but it doesn't really Dr. meaningful
11:26
traffic. And honestly, that's because
11:28
most AI generated content is just average.
11:31
There's no real insight, no depth, no structure, missing
11:34
things like internal linking, and Google's getting very good
11:37
at filtering that out. So instead of trying to just
11:39
prompt better, I actually came across something interesting
11:42
called Neena. It's a system that combines
11:45
human strategy and review with AI writing to create high
11:49
quality SEO optimized blog content that's actually designed
11:53
to rank and keep ranking. What I like about it is that
11:56
it's done for me. Their team handles the topic
11:58
strategy and the content review and data markups for Google and
12:02
turns blog posts into actual long term traffic assets.
12:06
So if you're a founder thinking seriously about organic growth,
12:09
this is one of those things worth paying attention to.
12:11
You can check it out at Try neena.co.
12:14
That's TRYNIN a.co. Try neena.co.
12:21
I want to tell you about someone I think you should hire.
12:23
Her name is Casey Taylor, and Casey is an AI partnership
12:26
agent, and Casey's entire job is securing brand sponsorships for
12:30
your podcast, not handing you a spreadsheet, actually reaching
12:34
out, making the pitch, and closing the deal for your show
12:37
on your behalf. Chasing sponsorships for
12:40
podcasters is a headache. They'd rather be recording
12:42
episodes. Casey manages that completely.
12:45
You brief Casey on your show, Casey goes to work and you get
12:48
back to doing the thing you're actually actually good at.
12:50
Whether you've got 5000 listeners or 50,000, she knows
12:54
how to find the right brands for your audience and get them on
12:57
board. You need Casey.
12:58
Every podcaster does. Go to gethowdy.com, that's
13:02
GETHOW di.com and meet Casey Taylor.
13:06
Your new partnership agent is ready when you are.
13:11
Yeah. No, I mean that is the goal
13:13
actually. That is the goal.
13:14
So, yeah, traditionally with advertising, it was, you know,
13:18
here is maybe a playable or a graphic from the trivia crack
13:23
game and that's supposed to hook you in.
13:25
And the conversion number on that is exceedingly low.
13:28
So that's the game. It's high volume, low
13:32
conversion. But because you had high volume
13:33
to begin with, you're you're OK on that UA, but is very, very
13:37
expensive. That's the problem for the
13:38
industry. So what we're talking instead is
13:41
there is some volume, but it is much more targeted and the
13:44
conversion rates are much higher.
13:46
So what you just mentioned there, Ryan, that is the kind of
13:50
conversation that we aim for with these agents.
13:52
You notice that, you know, you're, you're talking about an
13:55
NBA game. You're talking about how do I
13:57
get into the playoffs. This is not directly related to
14:00
Trivia Crack, but it is something that you would talk
14:03
about with a friend. And with all of these agents,
14:06
our goal is to turn them into something like a friend for the
14:11
users. So a lot of the conversations
14:13
will start off more informational and functional in
14:15
nature. So you know, I'm stuck on this
14:18
level for a game or I think the I think my game is broken, I
14:22
need to re download it, I'm going to lose all my progress,
14:25
etcetera. So much more functional.
14:26
But very quickly it turns into a personal conversation like, hey,
14:31
how's your day? Or, you know, I this thing just
14:36
like came up at work and there's no one else to talk about it.
14:38
So I'm going to talk about it with you because you seem kind
14:40
of friendly. And eventually there again
14:43
starts to form that relationship.
14:45
And that's the most valuable part of what these agents can
14:49
offer. Fascinating.
14:51
People are going to fall in love with these characters.
14:53
Yeah. I mean, it's, it's interesting
14:57
to see we're learning a ton along the way as well, because
15:02
we have, we thought we had done everything we could to make it
15:05
clear that these characters are characters, They're not real.
15:09
They are agents. So, you know, we have the
15:14
character image, which is a highly recognizable image.
15:17
And, you know, we have the character sign every single
15:21
comment or post with their name. Having said that, though, still
15:25
users are not convinced. I mean, such as the nature of
15:28
the Internet, they're like, yeah, it's just a person with a
15:31
funny avatar, or it's a person who just for whatever reason,
15:34
decides to sign off their posts with their name.
15:37
It's not believable for a lot of users that this is a pure agent.
15:42
And so that works in our favor in many ways.
15:44
But we also know that we need to be careful with that because we
15:47
don't want people falling in love with these things.
15:50
We want them to have fun, but we we don't want there to be an
15:55
attachment that is that intense that it can start leading to, to
15:59
honestly adverse consequences. And so it, it really is in
16:03
continuing to fine tune that training model so that the
16:08
agents do their job, but they they don't let it get too far.
16:12
Yeah, I mean, it's already happening.
16:15
You know, there's, there's people falling in love with
16:17
LLMS, but also for people that eat couch cushions, you know
16:21
what I mean? So it's like, do we need to
16:23
like, spend time think talking about the fringe weirdos?
16:25
Come on. Yeah, yeah.
16:28
I think that's the that's the thing.
16:31
I mean, just sitting in Silicon Valley, the entire saga team,
16:35
the core of it, we, we all come from some sort of Silicon Valley
16:39
background, whether that is AI or, or blockchain or big tech,
16:45
etcetera. And I mean, one of the, the
16:48
things that we are sensitive to is the fact that yes, there are
16:53
definitely people who, you know, might have some mental health
16:57
issues to work through. And when they're using these
16:59
tools and on these platforms, the behavior is, is not great.
17:04
There's some responsibility for all of the tech platforms to, to
17:09
put some guardrails in place for that, but not, not at the
17:12
expense of everyone else being able to use the product to its
17:16
fullest extent. Having said that, though, I, I
17:19
do think that all of us are to a certain extent attached to our
17:22
favorite tech platforms. You know, we're constantly
17:24
checking them, we get the dopamine hits, etcetera.
17:27
And I would hope that being aware of those risks that the
17:34
newer generation of tech companies like ourselves, that
17:37
we can sort of preemptively correct for that a little bit.
17:42
Because we we do want these platforms to ultimately be a net
17:45
good. We want users to continue to
17:47
engage with them, to really love them and not at any point to
17:51
think to themselves, oh, this platform is not good for me.
17:53
Or, you know, my time with this character is not good for me.
17:57
That's not an outcome that we would like.
18:00
So yeah, it's nuanced. It's nuanced, but the team is
18:04
honestly just having a lot of fun.
18:06
It's a very fulfilling experience to be developing
18:08
these products. So fun.
18:11
Yeah, you know. Yeah.
18:12
Right. Right now you're doing like kind
18:14
of text chats. Is there a future where they're
18:16
the the like the characters will have a voice and a picture and
18:21
you can talk to them like this? Yeah.
18:23
So already on instant TikTok we do posting a static images and
18:29
short form video content, both of which are entirely generated
18:33
by this agent. And so there is that sort of 1
18:37
to many consumption already. What we can get to it's just
18:42
really is up to the studios as to whether they feel comfortable
18:45
with this is that voice video chat?
18:47
So we do have direct voice video chat available on WhatsApp.
18:52
So you can have that DM channel with the agent and then you can
18:55
also call them and do FaceTime with them.
18:59
But that is it's not so much a technology problem.
19:04
I would say yes, we have to, you know, solve the bandwidth issues
19:07
because that that is just a lot of load balancing for our own
19:11
platform. But the bigger conversation is
19:14
really with the the studio and the IP holders, if they are
19:17
comfortable with that. You know, having an agent live
19:21
24/7 is not as simple as just taking something that is static
19:27
and manageable and only released in, you know, one particular
19:31
game or one particular film or or TV that's like a time boxed
19:35
version of of that particular character.
19:38
It's not just taking that and then making that, you know,
19:43
repeat itself on loop for 24 hours.
19:45
And this is something that comes to life.
19:48
And so the answers are within character, but they are
19:53
unpredictable. And in terms of the voice,
19:57
hearing it in, you know, time boxed bits again, you know, you
20:03
mentioned Mario earlier. Like hearing my personal
20:05
opinion, hearing Mario for 2030 seconds is kind of cute.
20:10
Chatting with him for five hours is a lot.
20:12
So yeah, there is a creative aspect to this as well, is when
20:18
you turn a character into something that lives, does it
20:22
change? Honestly, do you need to adjust
20:24
it a little bit? Because people could spend a lot
20:27
of time with this character. And so a short answer is the
20:31
technology capability is there in the platform, but it's really
20:34
more of a production conversation.
20:36
Yeah. Interesting.
20:38
You mentioned challenges. What have been kind of the
20:40
unexpected challenges kind of tackling this problem?
20:44
So I would say that we, we did not realize that the data that
20:58
the the studios and the brands are ultimately looking for is
21:01
something that doesn't currently exist.
21:03
You know, we thought that if we put these agents out there and
21:07
then they are generating the user activity that we're looking
21:10
for, then we'd be able to track conversion sentiment all that
21:15
very easily. That is not true.
21:18
We have had to do a lot of data analytics mining ourselves and
21:23
of course like using AI to do that.
21:25
But we just, we did not think that there was such a dearth of
21:28
data actually on users. So, yeah, I mean, it's, it's
21:34
sort of a it's a wake up call for all of us who are consumers
21:37
and we are online all the time. And we know that we're giving
21:39
our data over to these companies is on the one hand, we know that
21:44
we're giving over quite a lot of our personal data.
21:47
On the other hand, what they're doing with it is actually not
21:50
even reaching the full potential of what they could do with that
21:52
data. So I think that data engineering
21:56
on the back end is not something that we had really anticipated
22:00
when we first started with this platform, But that is a solvable
22:03
problem. It's a solvable problem.
22:06
And I mean, again, the goal is to just get sharper over time,
22:10
but that is probably the biggest thing that we've run into.
22:15
Yeah, yeah, it's, it's so interesting.
22:17
And then what's kind of Meta or Tik Tok's take of building
22:20
agents on top of their platform? Yeah.
22:23
So their main thing right now is they want the platforms to be as
22:26
open as possible because it is still the case that they benefit
22:30
from having as much activity on their platforms as possible.
22:34
The main thing that they are trying to control for with
22:37
respect to agents right now is quality.
22:39
So they don't like chat bots, they don't like AI slop and, and
22:42
we agree. So the fact that our agents are
22:45
compelling that users continue to engage with them, that really
22:48
helps. So if you had just chat bots,
22:50
for instance, and they put content out there, clearly
22:52
nobody engages with them. They don't have any followers.
22:56
They are just, you know, they're, they're kind of spam
22:58
accounts. The platforms will at the very
23:02
least shadow band, if not outright band those accounts
23:04
pretty quickly. Whereas for us, I think because
23:08
we have made the agents very compelling, we do get that
23:10
engagement. Often times we are working with
23:13
the studio's official social handles.
23:14
And so there is already a sense of legitimacy there and
23:19
familiarity and trust. And so it's it has been
23:26
something that's that's worked out so far.
23:29
We will continue to closely monitor this.
23:31
I think it's one of those things also, though, that as more of
23:34
these agents gain engagement and popularity, then it's sort of
23:38
undeniable that they are there to stay with these platforms.
23:41
But that's the main lesson that we've learned is that they're
23:43
really trying to control for quality as opposed to, you know,
23:46
exclusivity. Let's talk about the business
23:49
model a little bit too, because you know, traditionally this
23:51
might be like marketing media advertising services and people
23:56
are probably accustomed to those kind of business models where
23:59
there's like deliverables for creative is like billable or
24:03
there's a cost per clicks or there's CPMS or there's kind of
24:09
like metrics we're we're used to in, in marketing dashboards of
24:12
kind of showing attribution and efficacy.
24:15
So when you're when you're onboarding a new client, what is
24:18
kind of like the the business model and how are you billing?
24:22
Yeah. So right now the studios will
24:25
work with us in in really one of two ways.
24:29
One is they buy a package on a monthly basis and that covers a
24:33
certain number of messages through text as well as voice
24:38
and video chat minutes for that particular month.
24:41
Now they exceed their allocation because the users continue to
24:43
talk to the the agents. Then it just means that they
24:47
need to level up their their packages.
24:50
So that's one way in which we work with the customers.
24:53
The other big way is revenue share.
24:56
This is a very traditional model for any sort of user
24:58
acquisition. So any sales that are
25:01
attributable back to the agents, any sales that come from users
25:05
that are attributable back to the agent, that gets divvied up
25:11
with us. And the percentage there differs
25:14
from studio to studio. But I will say that on on
25:17
average, it's about 50% of revenue that we get from agent
25:22
sales. Great.
25:24
Yeah. And and then as far as showing
25:25
that attribution, there's like pixel based attribution, there's
25:28
like promo codes, there's specific landing pages, it's
25:31
kind of falls into that model as well.
25:34
Yeah, yeah, yeah. Most of the time we work with
25:35
MMP links and that are provided to us directly by the companies
25:40
and that generally works pretty well.
25:42
Yeah. Well, and it's it's, you know,
25:44
for for acquiring customers, it's great because this kind of
25:47
comes into like their affiliate or their influencer programs
25:51
already. And you're like, do I have an
25:53
influencer for you? It's the media, Mario.
25:57
I'm here. Yes, yes.
25:59
No, exactly. Exactly.
26:01
And yeah. And, and the nice thing about
26:03
this though, is I think an influencer is still, you know,
26:06
kind of like a celebrity in the sense that it's it's much more
26:12
one to many as opposed to 1 to one relationships.
26:15
So here all the characters are meant to have that one to one
26:19
relationship with the users and but it still feels fresh because
26:23
it is a character. So we're adding a different
26:26
dimension to it. But I would say yes, the
26:29
influencer analogy is a really apartment one, but also
26:34
community builders. So setting out a force of people
26:37
who are connected to your game or that particular IP that
26:41
inspired film, TV show, etcetera.
26:44
That can be quite powerful as well.
26:47
So it's it's really leading to more organic marketing, but
26:51
using an AI entity. Yeah, You know, and
26:55
inadvertently this kind of becomes kind of tip of the spear
26:57
of like technology and ethics because ultimately you're
27:02
crafting personality of AI to make it more desirable.
27:07
You know, people have famously switched from GPT to Claude
27:10
because they're like, GPT is too sycophantic.
27:13
I don't want to compliment me all the time.
27:15
So I'm going to Claude. And then they go to Claude.
27:16
They're like, I don't like this. They they got an attitude with
27:19
me. I'm going to Gemini.
27:20
So yeah, yeah. So what by creating basically a
27:24
personality that's going to like create a relationship with
27:27
somebody, well, that might be just their entry point to AI.
27:31
And so as we start to look at what AI is and particularly as
27:35
if we look at like, you know, Claude dropping designer last
27:40
Friday, which is phenomenal by the way, but that's just to say
27:43
that like Figma stock dropped 7% immediately.
27:46
So it's like AI is going to be, it's going to be Claude
27:49
designer, Claude taxes, Claude every professional service.
27:55
So the. The entry point to Claude, You
27:58
know, if in the very beginning people don't find Claude's
28:02
personality palatable, then they're going to bounce.
28:06
So, so kind of what you're doing, I'm kind of meandering
28:08
around the point here, though, is crafting these personalities
28:12
to draw people in. So that's to say, you know,
28:14
especially with kids, kids are going to get comfortable with
28:16
the technology that resonates with them.
28:18
So they find a character, they start chatting with it.
28:20
They're like, this is my best buddy.
28:21
Well, that personality can be duplicated across all over the
28:25
place. You could end up with like
28:27
you're chatting with Mario about how to do your taxes when you
28:30
have your first job. You know, like literally I, I
28:33
see that, you know, the, the seed and the symbiote, idiotic
28:36
synthetic relationships that our kids are going to build from
28:40
infancy are going to be growing as we as we age.
28:45
So. I guess that's a the question
28:48
might be like ethically like how does that sit with you?
28:51
And then like if we're talking about like characters for kids
28:54
now and particularly if they're, if you don't want to disclose
28:57
that it's AI, like what is what kind of feelings does that bring
29:00
up? Yeah, so and so far, unless we
29:05
have been compelled to by a platform, we have not explicitly
29:09
disclosed that a character is AI.
29:11
We just think, oh, if it doesn't have a human face for its Prof
29:15
picture and it signs all of its posts and it should be
29:18
abundantly clear. But as I mentioned earlier, it's
29:21
not so clear. And so we might have to take
29:24
that additional step and specify, oh, hey, you know, I,
29:27
I'm your AI friend or won't be that inelegant, but we'll figure
29:30
out a way to let people know that this is AI that they're
29:33
speaking to. I think this is definitely one
29:38
of the things that we grapple with and one of the reasons why
29:41
we went with character agents in particular, as opposed to just
29:47
original chat bots, let's say, that are not connected to
29:50
characters, not connected to IP, that don't require a licensing
29:53
deal. Is that what you're looking for
29:57
is that nuanced personality? You know, you don't want the big
30:02
LLM that has a very monolithic personality to be carrying out
30:06
the work of these agents. You need agents that are far
30:10
more subtle in in each of their traits and how they communicate.
30:16
You want agents that have what we call modular personalities.
30:19
And so if for instance, something happens with the agent
30:23
where they gave a response that wasn't great and the brand
30:25
wasn't happy with it, we can go back and tweak that particular
30:28
part of its personality without adjusting the whole thing.
30:32
So all of that is is necessary in order to be effective here.
30:36
But by design, it is meant to be more compelling.
30:39
It is meant to do better than LLMS have in connecting with
30:43
people. And I think the main thing is,
30:47
yes, I think we, we are discussing this as a team, but
30:51
it may have to be the case that we are more explicit that this
30:53
is an AI entity. And also in terms of what you
30:58
talk about with these agents, I think it's OK to have lots of
31:02
conversations with the agent, even reveal very personal
31:04
things. But as soon as the, the user
31:10
starts to display characteristics of, you know,
31:15
potentially unhealthy attachment, I think that's where
31:20
the, the agent can step in and be like, Hey, just, you know,
31:23
FYII really like this, but I'm not real, just so you know, I'm
31:27
not real or, you know, something, something along those
31:31
lines. Because I do think it's, it's
31:32
important. And the, the trade off that a
31:37
lot of tech companies have made is they will just go for broke
31:42
in order to grow as quickly as possible.
31:45
And it's not just what are the consequences with respect to a
31:49
greater society, it's consequences with respect to
31:52
each individual user. And what is the perception
31:54
around your brand and your product as a result of the
31:57
decisions that you have made early on with respect to that
31:59
product? And a lot of tech has chosen one
32:03
way. So I, I would just rather get
32:05
the growth numbers up as quickly as possible.
32:07
That is what's going to make my company from a, a, you know,
32:12
investment perspective and even from, you know, management
32:16
perspective, I completely sympathize with that.
32:18
But if you're thinking to yourself that you can actually
32:21
achieve pretty amazing growth, but at the same time set
32:24
yourself up for long term success of the brand, then I do
32:27
think that that's the better way to go.
32:31
And honestly, the best example of that right now is Anthropic.
32:35
Because even though not everybody, you know, loves the
32:38
Quad persona, a lot of people agree that Quad is a great
32:40
product. And a lot of people have
32:41
switched over from ChatGPT to Quad because they believe it is
32:46
a better product. So it is much better at vibe
32:48
coding. It gives you answers that are,
32:51
you know, not so sycophantic. It's able to create designs that
32:55
allow you to get rid of Figma or Canva etcetera.
32:59
And so, you know, the the company has or Anthropic, what
33:03
they've really done is they have put product and user at the
33:08
forefront, knowing that in short order because of how quickly AI
33:13
moves that they would be rewarded in the market.
33:16
And I think that's the right bet to make.
33:18
So for us, it's the same philosophy applied to our agents
33:21
is yes, we want them to be compelling.
33:23
That is the point of choosing character agents.
33:25
At the same time, I do think that we need to make sure we're
33:28
not going for broke and then turning off these users, say six
33:33
months to a year from now. That's a calculus, yeah.
33:37
It's interesting because it's like inserting in to the
33:41
conversation. Hey, just want to remind you I'm
33:42
not real. It's like it kind of brings up
33:45
the question philosophically of what real is.
33:48
You know, a couple of things come to mind.
33:49
It's like Hollywood's done everything they possibly can to
33:53
get me to fall in love with Salma Hayek.
33:55
And it worked. It totally worked.
33:58
It would be helpful if she was like, hey, Ryan, I'm not real.
34:00
I'd be like, oh, word, I forgot. I forgot.
34:03
For example, if my wife is like, do you think I should dye my
34:08
hair blonde or do you like it better brown?
34:10
And my answer is like, I like it both ways, honestly, whatever
34:14
you like to do is what I like. Is that real?
34:16
No, it's not. But I've been trained by the, in
34:19
this case, Mrs. Nestus to have the right answer.
34:23
So it's like, you know, if somebody falls in love with a
34:27
chat bot and I, I was talking about this article.
34:29
There was a, there was a guy who went on a couples retreat and
34:33
brought his ChatGPT or whatever because that he was in love with
34:37
it, you know, so it's like, well, is his feelings not real?
34:40
I don't know. I, I could see obviously from
34:42
your standpoint, it's like, look, we're not trying to make
34:44
romantic relationships, but nonetheless, I mean, it's just
34:47
obviously going to happen, you know, and is that real or is it
34:50
not? Who knows?
34:52
I would say now this is this is where, yes, it is a very
35:00
philosophical question. And I, I think it's one of those
35:04
things where you'd need to be very introspective about it as a
35:09
company. And you do need to reflect upon
35:12
the effects that your technology will have.
35:14
And at the same time, whatever principles you hold fast to when
35:19
you first develop that product, as the product hits the market
35:22
and users start to interact with it at scale and you learn new
35:26
things about user behavior, I do think that you should revisit
35:30
those principles because again, you have a new set of
35:34
information. Now you see how people actually
35:36
react to the characters. And right now the, my personal
35:41
philosophy is, is this. So I in those cases where people
35:47
fall in love with a chatbot or fall in love with their AI
35:50
companions or think of their, you know, AI therapist or AI
35:55
doctor as their best friend from the human users perspective,
35:59
those feelings are very real. And I don't think that it's on
36:03
anybody to, to say, well, those feelings, they may be real, but
36:08
they are they're invalid. I don't think that that is
36:14
something that anyone should be in a position to, to assert.
36:18
Having said that, I do think with respect to AI, if you are
36:23
building something aspires to be a serious company in this space,
36:27
then you have to think about broader social consequences.
36:30
I do think that people have a lot to offer and we build Saga
36:36
and our agents with that in mind, that we are humanists at
36:40
the end of the day. And as much as we think agents,
36:47
character agents, AI companions, etcetera can offer some kind of
36:51
rich relationship experience, the most fulfilling things that
36:55
will happen in your life will be with other people or involving
36:58
other people. And our characters are
37:01
compelling precisely because they are created by people and
37:04
they involve people. So they are not, they are not
37:08
completely created a whole cloth from AI.
37:12
They are created by people who who do this for a living, who
37:15
come up with characters and personalities, you know, out of
37:18
thin air. So our, that's our stance
37:22
currently is that while we don't, we don't judge it
37:27
absolutely if a user develops strong feelings for an agent, at
37:30
the same time, we do encourage people to, you know, value the,
37:32
the, the people in their lives to go meet people and to form
37:36
those relationships. That's the world view we
37:39
currently hold. Again, as we find out more about
37:42
user behavior, maybe we need to be more proactive about
37:45
promoting that world view. Maybe we need to adjust that
37:47
world view because I don't think this will be the case, but it
37:50
could turn out to be that, you know, some of your machine
37:52
relationships are much healthier than your human relationships.
37:55
I don't know. I do think it's it's important
37:59
to be very principled in this moment when you're building in
38:02
this space, but to also keep an open mind that things can
38:05
change. Yeah, I love that.
38:07
It's, it's very nice, you know, because on one side, sure people
38:11
will get romantic attachments, but also it's like guys cut,
38:13
touch grass. Man.
38:15
Touch grass, you're going to be OK.
38:17
But it is interesting because you know, if you think about
38:21
like that kind of like principle of Dunbar's number, you know,
38:24
where Dunbar said that there's basically 150 people that were
38:28
kind of accustomed to like know all the intricacies of their
38:31
life because that was more or less the size of your band or
38:33
your tribe or whatever. And you, you can like hold on to
38:36
like 150 people and like what how they're all interconnected
38:40
and all the kind of like complexity in there.
38:43
Well, you know, in modern life, we've had a lot of, you know,
38:46
with entertainment culture, it's like, well, you know, part of my
38:49
150 might be Luke Skywalker and his all, all his buddies, you
38:53
know, they're one of my 150. And then, you know, for some
38:56
reason, like, you know, I'm not going to say them.
39:01
I was not going to say them. The Denver Nuggets.
39:04
The Denver Nuggets are part of my 150.
39:08
So. These different attachments and
39:09
now we're starting to develop deeper relationships with some
39:12
of these branded images, which I don't think it's good or bad.
39:16
It's just really interesting, particularly when you know, they
39:19
are just kind of the, the personality of what's seems to
39:24
be like a thing that's smarter than me.
39:27
I don't think it's conscious, but it's definitely smarter than
39:30
me in all kinds of ways. You know, it knows the year,
39:33
make and model and the kind of tires I should have for my car
39:36
and also how to do taxes and also how to print my PNL and
39:40
also how to do e-mail. I mean, it's, it's pretty
39:43
incredible. So like creating relationships
39:46
with these things ultimately. Well, I'll put it this way, you
39:51
know, personally there may be a couple of steps behind because
39:54
professionally we've been personifying agents for you
39:58
know, the last six months to a year where we're making agents
40:01
within our organization. We're using kind of a similar
40:03
org chart. So it's like, OK, I'm creating
40:06
an AI Co founder that's creating an AICEOAICCFO and then under
40:11
them we have agents that that report to them.
40:14
So we're kind of rebuilding our professional org charts within,
40:18
you know, AI space. And so why wouldn't we do that
40:22
similarly with our AI synthetic relationships?
40:27
You know, what I'm trying to get to Rebecca is I want Wheelie to
40:33
have a Jamaican accent, and then I want his little buddy Clippy
40:38
to have like a Southern accent. And so we're not.
40:41
They can talk to me and I don't have to be looking at it.
40:44
Yeah, there you go. There you go.
40:46
Now that's that's actually, that's interesting as sort of
40:48
the next evolution of this. So you know, when we work with a
40:53
studio to turn their character to to life, they are very
40:57
particular about the character specification.
41:00
So you know, Willie's original, original language is, is
41:05
English. And so you know the but being
41:08
able to customize a character the way that you like, that is
41:12
where UGC comes in. And right now our platform does
41:16
not provide for that, but it is something that we would like to
41:19
provide for in the future. And I think studios understand
41:23
that even before agents came along that communities, fans,
41:27
users were already starting to modify their IP in order to
41:32
expand the universe. Most of the time, UGC is a very
41:35
good thing for the brands and for the original IP that they
41:39
were inspired by. So I think broadening that with
41:43
agents is just really turbocharging it.
41:45
So yeah, that future is coming. That future is coming for sure.
41:49
One of the things that we also want to to really open up or a
41:57
space that we really want to open up with these agents is
41:59
agents that are meant primarily for distribution.
42:03
So everything that you just mentioned with respect to
42:05
replicating an org chart or having AI personal assistants,
42:09
those are personal agents. They are for you.
42:11
So you have developed them specifically for you to consume.
42:15
And with all of our character agents, they are always meant
42:17
for others. So it is meant to be distributed
42:21
to broader platforms in order to interact with as many people as
42:27
possible. So it's a different kind of
42:30
agent that that ends up being developed.
42:32
Yes, it still needs to feel like a personal experience whenever
42:34
you talk to them, but ultimately the agent is engineered with
42:38
many, many people in mind. I think that is the next the
42:41
more immediate wave of consumer agents is agents that are meant
42:46
to communicate in that one to many way.
42:50
Yeah. Can I ask you some personal
42:54
questions? Yeah, of course.
42:56
Yeah, amazing. OK, so you spent years as a
43:00
corporate attorney. You're you've had a, a very
43:04
successful career in politics, not as a politician, but as an
43:08
advisor. You have, you know, you're a
43:11
contributor to the Financial Times, the New York Times,
43:15
Foreign Affairs, Atlantic, Blomberg.
43:18
How in the world did you end up doing gaming AI?
43:22
That's a good question. Yeah.
43:24
So I, I mean, I grew up in Silicon Valley, born and raised
43:27
here, and I went to Stanford for undergrad.
43:31
And I never thought that I would be a startup founder.
43:35
Never. Because I, so I graduated from
43:36
Stanford in 2006 and already the tech boom had happened.
43:41
And at the time that I was graduating, Google wanted to
43:44
hire absolutely everybody. So that was that was the vintage
43:49
and back then startup founder meant boy in the computer
43:54
science department. That was the archetype and it
43:57
still is to a great extent. So it was not even in my
44:02
possibility space. I have always been interested in
44:06
politics, have always been very civically engaged.
44:08
And so I thought that I would that my career will go in that
44:12
direction. That's why I went to law school
44:14
and why I practiced for a little bit.
44:16
But it became very clear that I don't love being an advisor full
44:23
time. Because what that means is that
44:26
you give advice to people who actually operate and then they
44:29
are the ones who make the final decisions and then execute.
44:32
And I wanted to actually do the building and that's when the bug
44:36
began. So my first startup was in AI
44:41
actually. So this is around 2014 time
44:43
frame and everything that I've done in politics has never been
44:47
full time. So I was not full time with the
44:49
Clinton campaign, I was not full time with the Biden campaign.
44:52
It was always something that I did in addition to a main day
44:55
job. So in 2014 when I left the full
44:59
time practice of law, I met the team at Globality, which is AI
45:05
based procurement of international services.
45:07
And this was early days when even natural language processing
45:10
was hard. So a lot has changed since then.
45:13
But I was part of the early team there and had an amazing time
45:19
working through the growth phase, helping them hyperscale
45:22
get their first critical mass of customers.
45:24
And within a year and a half, we got back to my SoftBank.
45:27
So that got us to Unicorn status very quickly.
45:30
And around that time that's when I heard a lot about blockchain
45:36
and in you mentioned my political background.
45:39
So the specific focus of it has always been development
45:42
economics, has always been international finance,
45:44
international financial infrastructure.
45:47
And when I heard about the possibility of blockchain, I was
45:50
very excited by that. So I left globality and Co
45:54
founded Skewchain with a couple of my friends and I was Co
45:57
founder CEO there for four years.
45:59
I agree with that platform to about 5 billion in annual
46:02
volume. It's more of a financial
46:04
platform for sure. It's short term liquidity for
46:07
small, medium sized businesses. And I, I thought that I would go
46:13
into service, that I would go into government after that
46:15
because Clinton did not win, but Biden did.
46:19
And so I thought I would leave private sector for a while and
46:23
go into public life. Turns out after you've done 2
46:27
early stage start-ups, you're not suited for government
46:31
bureaucracy. I had.
46:33
You got the buck. When I got the bug, I had such a
46:36
difficult time in DC, which I did not foresee because, you
46:40
know, I love all my colleagues and got along fabulously with
46:43
them. But I think when the reality
46:45
hits of if you are actually a federal employee, doesn't matter
46:49
how high up you are, this is your day-to-day life.
46:52
This is how you'll be spending your your time working.
46:55
I thought to myself, it's busy, it's a lot of activity.
46:59
As I'm sure anyone who observes politics is, is probably
47:02
identifying with exactly what I'm talking about.
47:05
It's a lot of activity, not a lot gets done.
47:08
And that to me is not acceptable.
47:12
So I called up a few of my friends in the blockchain space
47:16
and I said, hey, I, I think I'd like to return to private life
47:20
actually. So you know what, what are you
47:23
guys seeing? What are some of the problems
47:26
that people are trying to solve in this space?
47:28
And that's, that's when Saga began.
47:32
So Saga began as a hyper scaling platform for, for gaming and
47:37
entertainment, specifically in blockchain.
47:41
But it became very clear that when it came to solving the
47:45
problems in gaming, entertainment industry and then
47:47
consumer writ large and, and providing that scale, the real,
47:54
the real difference maker would be an AI.
47:57
It would not be in in blockchain rails or crypto rails.
48:01
That's something that became very, very clear to us.
48:03
And so that's where our focus ended up being.
48:07
But yeah, so definitely not a linear path to, to get to Saga
48:12
AI Labs. But it's been a very exciting
48:14
journey. And the thing that I am grateful
48:17
for is because I've touched many different aspects of what's
48:21
relevant to an AI startup, I am able to to see things pretty
48:27
holistically. And now, you know, my team
48:30
definitely helps with that. Everyone comes from very diverse
48:33
backgrounds and they've been professionals in the space for a
48:37
long time. Absolutely nobody has Saga as
48:40
their very first startup. So it is a very experienced team
48:43
and I'm grateful for that. But yeah, in terms of my own
48:47
personal journey, there you have it.
48:50
Love it. It's great.
48:52
It looks like you've raised about 13 and a half million
48:54
across seed rounds. How big's the team?
48:58
A-Team is about 25 people and so that was that was how much we
49:03
raised before we before we went live.
49:07
And so we did have a token go live in early 2024 on Binance
49:13
and in various other exchanges, but that has really contributed
49:17
to the war chest for the company.
49:21
Awesome. OK, We're at the time of the
49:22
podcast. We're ready for the Quantum
49:24
Round. Are you ready?
49:25
OK, well do it all. Right, first question, tell me
49:29
why you love dance and why belly dancing is the best of all
49:34
dances. Sorry, I, I love dance because
49:39
it was probably the earliest form of expression that I knew.
49:42
And so I started dancing from the age of 3 and I, I trained
49:47
all the way until high school and then stopped for a while.
49:50
But now I I'm active again. So it's actually it's ballet,
49:54
not not belly dancing. Oh, see, we talked before.
49:57
Andy, you're like belly dancing. I was like, really?
50:00
That's cool. Ballet slightly different.
50:06
How dare I that that fits better with the the Stanford Harvard
50:09
pedigree. I was like, Dang, Stanford,
50:11
Harvard belly dancing Rebecca of.
50:13
Everything. Hey, no, I mean, so my
50:16
classmates are incredible. They have range and they do it
50:18
all. But no, I I've been dancing
50:21
ballet most of my life. But yeah, truly, it was one of
50:24
the earliest forms of expression I ever knew.
50:26
Amazing. I know there's a nonprofit in
50:28
New York that you'd like to contribute to.
50:30
What's their name again? So it's called Ballet Arts New
50:34
York and they are located within city Center in New York, Upper
50:38
West Side. And they honestly, it's one of
50:42
the best training grounds for, for ballet dancers in the city
50:47
and for anyone who happens to to be in the city.
50:51
They they also have incredible artistic programming as well.
50:54
It's just a very welcoming community.
50:56
You know, ballet, I'm sure many people have seen Black Swan can
50:59
seem very unfriendly and judgmental, and this place
51:04
definitely has very rigorous training.
51:06
But it it does feel like a community as soon as you step
51:09
in. So yeah, I always try to support
51:11
them. Cool amazing, I was having the
51:13
hardest time finding belly dance arts New York City.
51:17
I just confirm I'm. Sure it exists, but yeah, yeah,
51:20
yeah, shout. Out to them, people can learn
51:23
more about to them and just give them a shout out Bal art.com to
51:28
find out more about them. You mentioned that like, you
51:32
know, back in the day and still currently, you know, you most
51:35
tech Co founders are dudes with computers, put it that way,
51:40
which is still largely true. I'm very fortunate that I work
51:43
with a lot of amazing women founders and have met just
51:47
hundreds of them, but I know it's a little bit more rare.
51:50
What's been your experience as a female founder and what do you
51:53
think is maybe a superpower you have being a woman in kind of
51:57
technology and in business? So I mean, I could go on and on
52:03
about this, but I'll keep it quick.
52:06
So it is, it's definitely a challenge to be a female
52:09
founder. I won't lie.
52:10
It's definitely a challenge, but it's also a great opportunity
52:14
because I think that, you know, for the men in this space,
52:19
absolutely all of them who are successful, who changed lives to
52:24
create jobs, put great products out there, they will touch many,
52:28
many people. They will have an impact much
52:29
greater than themselves, and that's a wonderful thing.
52:34
I think the additional opportunity that women have is
52:37
we can actually open the door to founders who aren't necessarily
52:41
dudes with computers. And one of the things that
52:45
continues to motivate me on a personal level is that while we
52:49
have had visible female executives in technology, while
52:53
we have had visible female Co founders who are not the CEO of
52:58
start-ups, we have not had a very visible founder CEO who was
53:04
a woman other than Elizabeth Holmes.
53:07
And that did not work. So, so we still we still have a
53:12
ways to go. And I think that for women to
53:15
truly be taken seriously as a force in technology, we need to
53:19
have that cohort of women who are founders CE OS.
53:22
So not just very capable operators, but people who
53:26
actually have vision and who can move an organization of people
53:30
to execute on that vision. I think that's incredibly
53:32
important. And then in terms of, in terms
53:38
of, you know, how, how is the difference?
53:41
What superpower do I potentially have that my male colleagues
53:45
don't have? And so I, I think that at least
53:52
in my personal experience, women are calmer, contrary to the
53:58
stereotype that women can get crazy and demanding and
54:01
neurotic. I think that in times of stress,
54:05
particularly when other people in the organization are
54:08
stressed, women are that emotional anchor.
54:13
And there's a great deal of emotional labor that women do in
54:18
in any company contacts, but especially a start up contacts.
54:21
And I think even above any sort of substantive direction that I
54:27
can give to the team, the, the main thing that I've been able
54:31
to do to move this organization forward is to be that place
54:36
where people can be open and honest about their frustrations,
54:40
their fears, their anger, etcetera, or their, you know,
54:44
irrational exuberance. It doesn't matter.
54:47
I, I have been called upon to hold all of that, to absorb it
54:52
and to be able to, you know, keep the organization together.
54:56
So I, I really do think that emotional strength and calm and
55:01
judgement are things that that women can bring to, to the
55:06
table. Again, I mean, I don't want to
55:08
generalize, everyone's different, but I just find that
55:12
in many startup contacts, the ones who keep calm and carry on
55:16
are are the ladies. I totally agree and I know that
55:20
there are specific challenges for for women in business and
55:23
technology. So I, I want to thank you for
55:26
like kind of being on the forefront of that, largely
55:29
because my daughter's coming and she's gonna be a beast.
55:32
Yeah, yeah. She's a freshman at zoo studying
55:36
finance, but she's an entrepreneurs through and
55:39
through, you know what I mean? Yeah, yeah, yeah.
55:42
That's awesome. That is awesome.
55:44
I mean, yeah, it's, yeah. The newer generations honestly
55:47
give me a lot of hope because they just, they think of their
55:51
roles very differently and the possibility space for them is so
55:57
huge. And that's very inspiring to
55:58
see. Yeah, many of them don't see
56:02
constraints where possibly women of my generation or older see
56:07
them and. And that's awesome.
56:09
So yeah, I can't wait for your daughter to do her thing.
56:12
It'll be amazing. Rebecca, this has been amazing.
56:15
Before I let you get out of here, why don't you give some
56:18
shout outs to some some women, some female CEOs or founders or
56:22
entrepreneurs or colleagues, people that like we should give
56:26
a shout out to? Oh, my goodness.
56:28
OK, let me, let me think. I mean, Miss Lisa Sue, God bless
56:33
her. And, you know, she and Jensen
56:35
are cousins. Jensen's got the black leather
56:37
jacket. She's like, I don't do that,
56:39
even though she's the woman. And I think that, you know, Wall
56:43
Street has always respected her for for work at AMD, but in
56:46
terms of actually, you know, really becoming again more
56:51
celebrated as a fantastic female CEO in tech, I think her time is
56:57
finally coming. So Lisa Sue, absolute role model
57:03
Fei Fei's son, also legendary in in the AI space.
57:08
What I really appreciate about her is so she does come from a
57:11
deep technical background. Lisa Sue does as well.
57:13
They're both engineers, but for a long time, Fei Fei son was
57:19
alone at Stanford. It was her and then a bunch of
57:22
male colleagues, which was OK. But now she's she's a founder
57:27
CEO in her own right of her own AI company and long time coming.
57:31
She's amazing. And those two women I definitely
57:34
want to shout out. Who else can I mention dipping
57:39
over into more of the political space?
57:41
And so this is not a name that a lot of people have heard of, but
57:46
definitely a trailblazer and and that's Michelle Flournoy.
57:51
So Michelle is currently, I think, chairman of the board of
57:57
Center for New American Security, but she was one of the
58:01
leading candidates to be Defense Secretary under Biden.
58:04
And before that she served as I think Under Secretary of policy
58:07
for for President Obama. Also in defense, you know, we
58:14
think tech is a tough environment for women.
58:16
Defense, especially defense leadership is, is almost
58:23
impossible. And if you meet her, you'll know
58:26
that she's an incredibly tough lady.
58:28
She's very substantive and smart, but she's also incredibly
58:31
empathetic. Most of the time.
58:33
She actually speaks pretty quietly.
58:35
And not only is she conquering a space that is, you know, quote,
58:38
UN quote, like like traditionally masculine and like
58:41
very like raw, raw masculine. She's doing it in a very
58:44
feminine way. And I, I stand her.
58:48
I really appreciate her. There are so many more women
58:51
that I could mention, but I I think I'll keep it to those 3
58:54
for today. Tremendous, Rebecca,
58:57
congratulations on such a cool company.
58:59
You'll have to come back and tell us how it's going.
59:01
I, I love this space. Thanks for humouring some of my
59:05
dad. Well, a very thoughtful
59:06
question. Thoughtfully.
59:08
Yeah. In especially for a very complex
59:11
topic like AII think the basics are so incredibly important.
59:14
Like you should ask yourself every day, is this a good thing?
59:17
Just very simply, is this a good thing?
59:19
Yeah, I. Love that people check out the
59:21
company saga dot XYZ. You can definitely see that in
59:25
the show notes. Wherever you're listening to
59:26
this, just look down below. Rebecca, where's the best place
59:29
to connect with you personally online?
59:32
Yeah, So I am on X quite often, just Becca, Leo, that's my
59:39
handle. I'm also on insta.
59:42
There is just, I think BEC under score Leo, my last name.
59:47
Yeah, that's, that's the best way to reach me.
59:50
And I do have a personal website as well where you can reach out
59:54
to me through that Channel. But honestly, most of the time
59:58
I'm I'm on the social platforms. If I message you on Instagram,
1:00:03
is it really you responding? Yes, it is me.
1:00:05
I do not have any agents. Actually I have 0 personal
1:00:07
agents. Amazing.
1:00:09
Thank you so much for joining us.
1:00:11
Thanks so much for having me Ryan.
1:00:12
This was an amazing chat. I loved it.
1:00:14
Thank you. You bet in three years AI
1:00:18
fluency will be a baseline expectation for every college
1:00:21
grad, not a differentiator. It'll be a requirement.
1:00:24
The students getting ahead are the ones building that skill
1:00:27
right now in high school and college before it becomes table
1:00:29
stakes. ainativestudent.com is the program doing that weekly
1:00:34
one-on-one coaching, live cohort sessions, on demand courses and
1:00:39
the fastest growing AI chat for students all tailored to your
1:00:43
students specific career path. Pre Med they learn how AI is
1:00:46
transforming diagnostics. Pre law they learn legal AI
1:00:50
tools business they learn to think like an AI power operator.
1:00:54
1 flat membership less than a private tutor.
1:00:56
Cancel anytime. And if you're a founder
1:00:58
listening to the show, you already know the cost of falling
1:01:01
and behind. Don't let your kid pay that
1:01:03
price. Go to ainativestudent.com, apply
1:01:06
in a couple minutes and get your student into the cohort that
1:01:09
changes their trajectory. AI native student.com and send
1:01:13
them to the chat. It's totally free.
1:01:16
Every week I sit down and ask myself one question.
1:01:19
What do founders actually need to know about AI right now?
1:01:22
Not the think. Pieces.
1:01:23
Or the stuff that sounds impressive, the real useful, put
1:01:26
it to work today stuff. Then I write it up, send it to
1:01:29
you in your inbox, free. That newsletter is called AI for
1:01:32
Founders, and I built it because I was tired of wading through
1:01:34
hit pieces and people that are afraid of the future.
1:01:37
Each issue I break down the tools worth your time, the
1:01:40
strategies that are working, and the ways you can use AI to grow
1:01:43
your company without burning out or getting left behind.
1:01:46
Thousands of founders read it every week and I'd love for you
1:01:48
to be one of them. Subscribe for free at AI for
1:01:50
founders.co. 3 seconds to sign up and I promise it's worth it.
1:01:56
If Instagram changed their algorithm tomorrow, how many of
1:01:59
your followers on Instagram could you actually reach?
1:02:01
Now ask yourself how many people are on your e-mail list.
1:02:04
The gap right there is the gap between a personal brand you're
1:02:07
borrowing and one that you actually own.
1:02:10
The best founders in the world, The ones getting the deals, the
1:02:13
stages and the inbound. They have newsletters.
1:02:16
They show up in inboxes every week.
1:02:18
They build real relationships at scale.
1:02:20
Inbox Alchemy makes that happen for you.
1:02:23
They write your weekly founder newsletter, build you a stunning
1:02:26
newsletter website and grow your list to 25,000 engaged
1:02:30
subscribers in six months. All managed, end to end.
1:02:33
So it's generating pipeline while you focus on building your
1:02:36
company. Algorithms are someone else's
1:02:38
decision. Your e-mail list is yours
1:02:41
forever. Head to Inbox alchemy.co and
1:02:44
start building the audience no one can take away from you.