Latest / Chief in Tech Deep Dive Shortcast / Chief in Tech S1E46 - Kalina Tonkovska : Transforming Legal Operations – The Impact of Automation and AI

Transcript

Chief in Tech S1E46 - Kalina Tonkovska : Transforming Legal Operations – The Impact of Automation and AI · Chief in Tech Deep Dive Shortcast

0:00
I want you to think for a second about the backbone
0:03
of literally every business interaction you have.
0:06
Oh, the stuff no one really wants to think about.
0:08
Right, exactly. But I mean, whether you are leading
0:11
a massive team or prepping for a board meeting
0:14
or just, you know, signing up for a new software
0:16
tool, there is one thing sitting at the center
0:19
of it all, the contract. Always the contract
0:21
always but there is this glaring paradox right
0:25
now If you look at how software has radically
0:28
transformed entire industries over the last decade
0:30
I mean we have algorithms predicting global supply
0:34
chain shifts in milliseconds Yeah, and self -driving
0:36
cars navigating city streets exactly yet The
0:40
very promises that make all of that business
0:42
happen, the contracts, they are treated like
0:44
ancient artifacts. They really are. They're just
0:46
snatic documents, buried in cloud folders, locked
0:49
away in PDFs, or completely lost in some endless
0:53
chain of emails. It is a striking contradiction,
0:56
honestly. We are operating in this hyper -fast
0:59
digitally native world, but the legal framework
1:02
holding it all together hasn't evolved much past,
1:05
well, the digital equivalent of paper in a filing
1:08
cabinet. A digital filing cabinet. Right. You
1:11
have this massive asymmetry between the speed
1:14
of a company's technology and the speed of its
1:16
legal agreements. Which is exactly what we are
1:18
tackling in today's Deep Dive. We are pulling
1:21
from a really fascinating session titled, Contracts
1:24
That Think. AI -driven operations. A great session.
1:28
Yeah, it was presented by Kalina Tunkowska. She's
1:31
the founder and CEO of a UK -based company called
1:34
Programmatic. And our mission today is to explore
1:37
this fundamental shift she outlines, taking these
1:40
passive dead documents and turning them into
1:43
living breathing systems. Which is a huge leap.
1:45
It is. We are going to look at how AI is bridging
1:49
the gap between a written promise and actual
1:53
automated business operations. And the stakes
1:55
here go far beyond just, you know, upgrading
1:57
a filing system. The shift toward what Tonkovska
2:00
calls programmatic agreements, it's about fundamentally
2:03
changing the nature of a business deal. Changing
2:05
the DNA of the deal itself. Exactly. We're moving
2:08
away from simply managing contracts to letting
2:11
the contracts manage the operations. Wow. OK.
2:14
But before we get into the mechanics of how an
2:17
AI -driven contract actually works, we really
2:20
need to understand the problem with the current
2:22
status quo. Right. The baseline. Yeah. We have
2:25
to look at how dead our current systems are to
2:28
appreciate why this shift is even necessary.
2:31
Because a contract is fundamentally defined as
2:33
a promise to make things happen. A promise to
2:36
make things happen. I like that. But right now,
2:38
instead of driving action, they create bottlenecks.
2:41
They are completely passive. It reminds me of
2:44
an old recipe card locked in a kitchen drawer.
2:47
Oh, that's a good way to look at it. Right. The
2:49
recipe tells you exactly what to do. It has all
2:51
the instructions, the exact measurements, the
2:53
temperature for the oven. But it is just sitting
2:56
there in the dark. Just waiting. Yeah. It can't
2:58
actually cook the meal for you. You still have
3:00
to source the ingredients, measure them out,
3:02
and do all the heavy lifting yourself. The recipe
3:04
card has all the logic. but zero execution capability.
3:08
That perfectly captures how traditional contracts
3:11
function. I mean, when you sign a standard vendor
3:13
agreement or a service level agreement, all you
3:18
have done is agree on the rules of engagement.
3:20
You've just written the recipe exactly you establish
3:23
the pricing the deliverables the compliance standards
3:26
the timelines But the moment the digital signature
3:28
is stamped that document just goes to sleep It
3:32
goes right into a subfolder on a hard drive never
3:34
to be seen again until I don't know someone threatens
3:37
a lawsuit or until someone misses a payment Yeah.
3:39
And this is the core operational roadblock, right?
3:43
Business moves at breakneck speed, but when the
3:45
promises that dictate how that business should
3:47
operate are locked inside a static PDF, the execution
3:51
of those promises slows to an absolute crawl.
3:54
Because humans have to step in. Yeah, humans
3:57
have to manually read the text, extract the obligations,
4:00
plug them into a project management tool, set
4:02
calendar reminders for renewals. The calendar
4:04
reminders. And constantly cross -reference spreadsheets
4:07
just to make sure they're remaining compliant.
4:09
The contract itself is doing absolutely zero
4:11
work to ensure its own promises are kept. Okay,
4:14
let's unpack this. Because if a static document
4:17
is the bottleneck, how exactly does technology
4:19
inject life into it? Donkowska describes these
4:23
programmatic documents as intelligent self -executing
4:27
agreements that combine natural language with
4:30
executable code. Yes. Now, I got to say, when
4:33
I hear executable code mixed with a legal agreement,
4:36
my immediate fear is the barrier to entry. Sure.
4:39
People get intimidated. Right. Does that mean
4:42
every business now needs to hire an army of software
4:45
developers just to sign a standard vendor agreement?
4:49
Do I need an engineer sitting next to my legal
4:51
team for every single deal? That is the trap
4:54
most people fall into when they hear the word
4:55
code. They immediately assume they need developers.
4:58
It sounds expensive. It does, but the reality
5:00
is the opposite. The current state of the technology
5:03
allows any business to elevate their contracts
5:06
into programmatic agreements by leveraging large
5:09
language models, or LLMs. You use those alongside
5:12
platforms that provide this specific infrastructure.
5:15
The AI is the thing doing the heavy lifting of
5:17
translation. Let me push back on that for a second,
5:19
though. If this AI is acting as a translation
5:22
layer, executing code based on legal text, what
5:25
happens if it hallucinates? Ah, the hallucination
5:29
question. Yeah! I mean, if the LLM misinterprets
5:32
a standard net -30 payment term as... I don't
5:35
know, net 300, does the system just blindly delay
5:38
my payments for 10 months? That is a critical
5:40
distinction to make about how this technology
5:42
is actually deployed. We aren't talking about
5:45
a generative AI chatbot just freely writing code
5:49
or making unguided decisions. So it's not just
5:52
chat GPT writing my contracts? No, not at all.
5:54
In a programmatic contract platform, the AI acts
5:57
as a highly structured extraction and mapping
6:00
layer. It reads the plain English, or whatever
6:02
natural language you are using, and it recognizes
6:05
specific conditions, dates, and deliverables.
6:07
Okay. And then it maps those variables into a
6:10
rigid, predefined logic framework. So it's not
6:12
writing the software from scratch. It's more
6:14
like filling out a digital logic puzzle based
6:16
on the contract's text. Think of it like a series
6:18
of if -then statements. If -then statements.
6:20
Got it. The AI translates the legal text into
6:23
structured logic, like if condition X is met,
6:27
then trigger action Y. Right. Once that logic
6:29
is established, the platform uses standard software
6:33
mechanisms, like APIs and webhooks, to communicate
6:36
with your existing tools. Can you give me an
6:38
example of that? Sure. Let's say the contract
6:41
stipulates a payment is due upon approval of
6:44
a deliverable. The AI doesn't hold your money.
6:47
Instead, the programmatic contract is integrated
6:50
directly with your project management software
6:52
and your accounting software. When the project
6:55
manager clicks approved in the management tool,
6:58
that system sends a signal to the contract. The
7:00
contract acknowledges the if condition has been
7:02
met and automatically pings your accounting software's
7:06
API to instantly issue the invoice. So the AI
7:09
is just the bridge that takes the plain text
7:11
and wires it into the actual digital nervous
7:13
system of the business? Exactly. The contract
7:15
is literally talking to the CRM and the accounting
7:18
software, which brings us to the real world impact.
7:22
So what does this all mean? Automating this agreement
7:25
logic obviously cuts down on operational expenses,
7:28
right? By eliminating manual tasks across administration,
7:32
project management, and account management. Massively.
7:35
But what does this actually look like on the
7:37
ground? If the API is firing off the invoices
7:39
and checking the deliverables, what happens to
7:41
the account managers? Are we basically putting
7:44
project managers and account managers on autopilot?
7:47
Are we just automating them out of a job? It's
7:50
a fair question, but no. We are just removing
7:52
the robotic tasks from the human's plate, which
7:55
actually elevates the role of the account manager.
7:57
It elevates how? Well, consider what an account
7:59
manager does on a daily basis in a traditional
8:01
setup. A massive portion of their week is spent
8:04
double -checking things. True. Lots of admin.
8:06
Yeah. Did the client hit the usage tier that
8:08
triggers a billing increase? Did we deliver the
8:11
localized assets by the date specified in section
8:15
four paragraph B of the master service agreement?
8:18
Digging through PDFs again. Exactly. A programmatic
8:21
contract handles that monitoring automatically.
8:25
It knows the rules, it actively monitors the
8:27
data feeds from the product, and it executes
8:29
the action, like sending the upgraded invoice
8:32
or flagging a missed deadline to the internal
8:34
team. So it's absorbing all that administrative
8:37
busy work so the humans can actually manage the
8:39
relationships and the strategy. Yes. They aren't
8:42
spending their Fridays buried in spreadsheets
8:44
trying to figure out who owes what. Which honestly
8:47
sounds like a dream. It is the definition of
8:49
moving at the speed of software. But beyond just
8:52
saving time on invoicing, the most significant
8:55
impact on the company's bottom line is the reduction
8:57
of legal risk. The risk factor. The ability to
9:00
automatically remain compliant with regulatory
9:02
changes is a massive shift. Let's explore that,
9:06
because the word compliance usually strikes fear
9:08
into the heart of literally any operations team.
9:11
Oh, absolutely. How does a thinking contract
9:13
handle a sudden regulatory change better than,
9:16
say, a human legal team? Well, think about the
9:20
traditional static process when a new regulation
9:23
drops. Say there's a major update to data privacy
9:27
laws like GDPR or a new state tax code. Panic
9:31
ensues. Total panic. Suddenly, a company has
9:34
to scramble. The legal department has to manually
9:36
review thousands of existing contracts to figure
9:39
out which vendors, clients, or partners are affected
9:42
by the new rule. Which takes forever. Forever.
9:44
Then they have to manually draft and send out
9:47
amendments. It is highly reactive, slow, and
9:50
incredibly prone to human error. You were basically
9:53
hoping a tired paralegal doesn't miss a crucial
9:55
clause in a 50 -page document. Exactly. Now consider
9:58
the programmatic approach. If your contracts
10:01
are structured as living logic, and they're integrated
10:03
with a system that monitors regulatory databases,
10:06
the contract itself can cross -reference its
10:08
own terms against the new law. Wait, really?
10:10
It checks itself? It checks itself. It can automatically
10:13
flag itself to the legal team, highlighting the
10:15
exact clause that is now non -compliant. In some
10:18
advanced setups, it can even halt specific automated
10:21
workflows, like data sharing with a third -party
10:23
vendor until the compliance issue is resolved.
10:26
Wow. It shifts the entire organization from a
10:30
reactive panicked legal strategy to a proactive
10:33
automated one. That's the beauty of it. That
10:35
is a fundamental paradigm shift for risk management.
10:39
But here's where it gets really interesting to
10:40
me, because it is one thing to automate basic
10:43
admin tasks and act as a compliance guardrail,
10:46
right? Right. The real leap happens when you
10:49
embed these contracts deeply into the company's
10:51
ecosystem to track performance. The concept of
10:55
applying adaptive learning for real -time actionable
10:57
insights. Yes, adaptive learning. If we go back
11:00
to that recipe card analogy we talked about earlier,
11:02
it's not just that the recipe is now cooking
11:05
the meal for you. It's like having an incredibly
11:08
observant data analyst baked right into the fine
11:11
print. I love that. It's watching how the business
11:13
performs, noticing that you always run out of
11:15
flour on Tuesdays, and then dynamically adjusting
11:18
the supply order for next week. What's fascinating
11:20
here is that term you used, adaptive learning.
11:22
That is the key differentiator. A traditional
11:24
contract only looks backward. Backward, how so?
11:27
It is a historical record of a promise made in
11:30
the past under a specific set of circumstances,
11:33
totally frozen in time. But a programmatic contract,
11:37
because it is embedded in the operational ecosystem,
11:41
is looking forward. Because it's a living system
11:43
absorbing performance data in real time. Exactly.
11:47
It isn't just executing a past promise. It is
11:50
actively learning from the execution of that
11:53
promise to generate insights for future decisions.
11:56
Do you have an example of that? Let's look at
11:58
vendor management. Suppose you have programmatic
12:01
contracts with 10 different logistics suppliers.
12:04
Because these contracts are intrinsically linked
12:07
to your supply chain tracking software, they
12:09
are constantly measuring their own performance
12:11
against their stated obligations. OK, tracking
12:14
the metrics. Right. And the system can aggregate
12:16
that data and tell you, you know, supplier A
12:19
consistently meets delivery deadlines 20 % faster
12:22
than supplier B, but supplier B has a significantly
12:24
lower rate of damaged goods. Oh, wow. So you
12:27
aren't waiting for a quarterly review to figure
12:29
that out. No, not at all. The contracts themselves
12:31
are bubbling up the intelligence. It turns the
12:34
legal department, which is traditionally viewed
12:36
as just a cost center, into a real -time data
12:39
engine for the operations team. Which perfectly
12:42
aligns with Kalina Tunkowska's entire thesis.
12:45
She didn't arrive at this concept purely from
12:47
a legal perspective. She arrived at it from an
12:49
operational one. That makes total sense. Her
12:51
track record involves scaling lean startups,
12:54
focusing on product innovation, and building
12:56
data -driven teams that move exceptionally fast.
12:59
When you're trying to scale quickly, anything
13:01
that doesn't actively contribute to momentum
13:03
is a liability. Dead data is a liability. And
13:07
a static PDF is the ultimate form of dead data.
13:10
Completely. From Tonkowska's perspective, efficiency
13:14
and automation aren't just cost -saving measures.
13:16
They're the baseline requirements for survival
13:18
in a digitally native market. Yeah, you have
13:20
to have it. And while this applies broadly to
13:23
all businesses, she specifically calls out three
13:26
industries that stand to gain a massive competitive
13:28
advantage by adopting AI -driven operations.
13:32
Real estate, finance, and healthcare. Okay, I
13:34
want to dig into those three because they are
13:37
heavy hitters. Why real estate, finance, and
13:41
healthcare specifically? What makes them the
13:43
perfect candidates for a complete overhaul of
13:46
how they handle agreements? It really comes down
13:48
to their unique vulnerabilities. These three
13:50
sectors share a very specific operational profile.
13:54
They are heavily regulated, they carry massive
13:57
administrative overhead, and they rely on incredibly
13:59
dense complex contracts to execute even the most
14:03
basic transactions. Yeah, they are notoriously
14:06
bogged down. Right. They're industries where
14:08
the friction of the static contract is the most
14:10
painful. Let's start with real estate then. That
14:13
is an industry practically built on towering
14:16
stacks of paper. Oh yeah. How does a programmatic
14:18
contract actually change the game there? Consider
14:21
the escrow process in commercial real estate.
14:24
It involves multiple parties, right? Buyers,
14:27
sellers, banks, title companies, local governments,
14:30
and a highly sequential list of dependencies.
14:32
Like a domino effect. Yes. Funds cannot be released
14:35
until the title is cleared and the title cannot
14:37
be cleared until the inspection contingency is
14:39
lifted. In a traditional setup, humans are manually
14:42
verifying each of these steps, sending emails,
14:45
and waiting for manual sign -offs. Which takes
14:47
weeks. Exactly. A programmatic contract wires
14:51
all of those dependencies together. Once the
14:54
digital title verification is confirmed via an
14:56
API from the local municipality and the inspector's
14:59
digital sign -off hits the system, the contract
15:02
automatically triggers the bank's API to release
15:05
the escrow funds. Wow. You eliminate days or
15:08
even weeks of administrative waiting. Instantly.
15:11
You take a process that usually requires a dozen
15:13
phone calls and compress it into an instant digital
15:16
handshake. That makes total sense. What about
15:18
finance? The stakes there are obviously tied
15:21
heavily to market volatility and compliance.
15:24
Finance is the perfect environment for programmatic
15:26
agreements because the data is already highly
15:28
structured. Take something like a margin loan
15:30
agreement where a client borrows money against
15:33
the value of their investment portfolio. Okay.
15:36
The contract stipulates that if the portfolio
15:38
value drops below a certain threshold, a margin
15:40
call is issued requiring the client to deposit
15:43
more funds. Right, a pretty standard setup. Right.
15:46
But a programmatic contract tied directly to
15:49
real -time market data feeds doesn't need a risk
15:52
manager to run a daily report. The moment the
15:54
market dips and the threshold is breached, the
15:57
contract executes the margin call automatically.
16:00
Immediately. It guarantees zero compliance drift
16:03
and completely removes the human latency that
16:05
can cost a financial institution millions in
16:08
a volatile market. It's enforcing its own terms
16:10
the exact second the conditions are met. Precisely.
16:13
OK. And finally, health care. Now, this one feels
16:16
the most complex because you are dealing with
16:18
patient data. privacy laws, and incredibly strict
16:21
ethical standards. In healthcare, the friction
16:24
is almost entirely about compliance and access
16:26
control. Hospitals deal with thousands of business
16:29
associate agreements, or BAAs, with third -party
16:31
software vendors, equipment suppliers, independent
16:34
practitioners. It's a huge web of contracts.
16:36
A massive web. And these agreements dictate exactly
16:38
who can access sensitive patient data and under
16:41
what circumstances. A programmatic contract can
16:44
tie those legal permissions directly to the hospital's
16:47
IT infrastructure. Wait, so the legal text actually
16:50
controls the firewall? In a sense, yes. If a
16:53
temporary contractor's agreement expires on a
16:55
Friday, the programmatic contract doesn't just
16:58
send an email to HR reminding them to do it.
17:01
It automatically communicates with the hospital's
17:03
active directory to revoke the contractor's access
17:06
to the patient database right at midnight. Oh,
17:08
that is brilliant. It ensures that the operational
17:11
reality of the hospital perfectly mirrors its
17:14
legal obligations at all times. For industries
17:17
like healthcare, finance, and real estate, moving
17:20
from passive documents to self -executing workflows
17:23
isn't just a nice upgrade. It is a structural
17:27
advantage that allows them to completely outmaneuver
17:29
competitors who are still stuck pushing paper.
17:32
They are operating at the speed of software while
17:34
their competitors are still digging through cloud
17:36
folders trying to find the right PDF. Exactly.
17:39
It is a phenomenal shift in how we think about
17:41
the basic building blocks of business. So if
17:44
you are leading a team right now, managing vendor
17:46
relationships, or just trying to navigate the
17:48
future of your industry, this is a crucial mental
17:52
model to adopt. Highly crucial. The fundamental
17:54
promises of business are no longer just words
17:57
on a page. Through the power of AI translating
18:00
natural language into executable logic, contracts
18:03
are becoming automated code. Moving at software
18:05
speed. Yeah. They are cutting operational overhead,
18:09
enforcing compliance proactively, and using adaptive
18:12
learning to generate real -time insights from
18:14
your daily operations. It is the evolution of
18:17
the contract from a defensive historical record
18:20
into a proactive strategic asset. By embedding
18:24
these living systems into the business ecosystem,
18:28
you aren't just protecting the company, you are
18:29
actively accelerating it. It completely reframes
18:32
the entire purpose of a legal agreement. But
18:35
I want to leave you with one final thought to
18:37
mull over, something that pushes this concept
18:40
of adaptive learning to its logical and maybe
18:43
slightly wild conclusion. Okay, let's hear it.
18:45
We've talked about programmatic contracts embedded
18:48
into a company's ecosystem acting in real time
18:50
based on live data. But what happens when two
18:53
of these AI -driven contracts representing competing
18:56
companies have to interact with each other? Right.
18:58
Imagine a scenario in a fast -moving supply chain
19:02
where two thinking contracts are authorized to
19:05
negotiate delivery terms based on live weather
19:07
data and fluctuating fuel prices. What happens
19:10
when they execute an adaptive split -second compromise
19:13
that neither human party originally anticipated
19:16
when they set the parameters? That is wild to
19:19
think about. The recipe card isn't just cooking
19:21
the meal anymore. You might just decide to invent
19:24
an entirely new dish while you aren't looking.
19:26
It introduces an entirely new layer to the concept
19:29
of automated business, especially regarding how
19:31
much autonomy we are actually willing to hand
19:33
over to the agreement itself. It definitely changes
19:35
the weight of what it means to hit. I agree.
19:39
Thanks for joining us on this deep dive.