PERSUIT Exchange: What the Full Day Surfaced

For In-House Legal Teams

Every in-house leader in the room is carrying two jobs: the legal work itself, and the increasingly visible job of orchestrating how that work gets bought and delivered. The day confirmed the second job is now the one the C-suite is watching.

Pricing pressure

The pressure you feel is structural, and it isn't yours alone

Rate increases were the loudest signal of the day. But the room made clear the pressure isn't coming from legal. It's coming from the CFO and CEO, as a mandate, not a preference. One CLO described walking into budget season with a multi-million-dollar increase and being told, flatly, that it had to change.

Your peers are navigating the identical squeeze. The value of a room like this is a shared language for what "value" actually means before you walk back into a firm review.

From the room"For every dollar of legal spend, the business has to generate many more in revenue. That math is what's driving the conversation now."

The repricing

The numbers back you up when you challenge a rate increase

Across the last few years, profit-per-equity-partner at the largest firms has risen sharply while billable hours have barely moved. The bulk of that growth is rate-driven, not productivity-driven. That's a gap that is hard to defend once it's named out loud.

You don't need to win the argument on principle. You need the data, the courage to test the next firm, and a culture that lets your team move work without fear of being second-guessed.

Insourcing

Bringing work in-house is now a live lever, and the AI investment has to earn its return

One CLO has rebalanced significant volume in-house, with internal lawyers doing work that used to sit with junior partners at a fraction of the cost. Another tied part of their own bonus to an AI-driven saving on external counsel. The throughline: when you invest in technology, the business expects a return, and that often means less work going out.

The discipline that makes it work starts with one question before you ever pick up the phone: do I actually need outside counsel for this?

Fixed fees

Predictability beats the hourly fight, when the scoping is real

The in-house preference across the panel was unanimous: fixed fees, properly built. Predictability is what lets you justify a budget to the business; margin anxiety is the firm's problem to solve, not yours. Where AFAs worked, the difference was always the same: genuine scoping up front, established trust, and an early warning when assumptions break.

From the room"Tell us at 75 or 85% of the estimate that it's going over. At 110%, there's very little I can do to reopen the budget."

AI transparency

Your AI bar is high, and most firms haven't cleared it

Three in four in-house respondents have only generic "review the AI output" guidance, with no workflow-specific protocols. Two in three say they need the full picture before they'll say yes: tools used, data inputs, human review, confidentiality controls, verification ownership, error handling, and pricing impact.

That's a demanding standard, and almost no cross-party conversation has come close to meeting it. If you haven't set it explicitly, you and your firms are both running on assumptions.

Joint governance

You believe AI governance should be shared, but almost no one has built the structure

When the room voted on who should own AI governance, the answer was overwhelmingly "joint." 83% believe the decision should be made together. Yet only a small fraction of relationships have an actual shared working model, and the single biggest blockers named were security and governance, both largely on the client side.

The framework exists to fix this. What's missing is the structured conversation, and that one is yours to start.

AI economics

The savings question is the one nobody has answered yet

Asked how AI efficiency gains should be split, the room scattered: negotiated case by case, or not addressed at all, far outweighed any shared model. The aspiration that came through clearly: efficiency should show up in fixed or outcome-based pricing, not as the same hourly bill with fewer hours buried inside it.

The teams moving fastest are experimenting now with phased fixed fees, hybrid models and shared-savings conversations, before this becomes a fight.

The profession

If AI absorbs the associate's work, who trains the next great lawyer?

The most-quoted line of the day, in various forms: the model is already better than a junior associate at a widening set of tasks. That raises a question bigger than any one matter. If AI does the work that used to develop young lawyers, where do the great ones come from in ten years?

Framed as shared responsibility rather than cost rebalancing, this is the conversation that moves a market: clients and firms protecting the apprenticeship that built both sides.

For Law Firms

The clients in the room weren't asking firms to cut their margins. They were asking to be treated as partners in the economic conversation, signaling, clearly, what now decides who stays on the panel and who quietly comes off it.

Panel ROI

Clients are doing the math on the panel, and the business of law is now a criterion

Panels have been cut hard: one team from 300 firms to 19, another from 160 to about 40. Strong legal work and great service are the price of entry, not the differentiator. What keeps you on the list now is the business of law: how you engage on the relationship, your tech stack, and what you're doing on AI and efficiency.

From the room"You may be a star performer. But if you're not engaging with me on the business of law, I can guarantee you won't be part of the panel going forward."

Commoditized pricing

The bigger shift behind the AI ask: a commoditized legal marketplace is forming

The panel's single AI ask sounded gentle — share some of the upside. But the structural shift underneath it is anything but. The delivery model itself is being commoditized. AI-assisted workflows, standardized inputs, and predictable scope are turning what was bespoke legal work into something closer to a productized service.

Fixed price isn't the play here — it's the symptom. The real move is structural: a commoditized legal marketplace where price reflects a known unit of delivery, not the hours spent producing it. Firms that get there first will write the new pricing language. The rest will be measured against it.

Transparency unlocks

Proactively sharing your AI approach is the unlock, not the risk

Clients are restricting firm AI use less because they distrust it and more because they lack the governance to approve it. You may be ready to deploy; your client may not have a framework to say yes. The firms winning on AI aren't the most advanced. They're the ones making it easiest for a client to trust them.

Show the tools, the safeguards, the quality checks, and the pricing impact before you're asked. Demonstrate rather than market it.

Trust signals

Shadow billing is a trust signal, not an admin complaint

Firms named it directly: a request for shadow billing or WIP under a fixed fee reads as a lack of trust. But clients ask for it precisely because they can't tell whether the fixed fee reflects real efficiency or just a negotiated number. Both sides are operating in an information vacuum.

The way out runs through transparency up front: on scope, on AI use, on how time is actually spent. Firms that can show their work at the outcome level earn the right to send one number.

The conversation

Clients don't expect the bill to drop. They expect the conversation to change

A telling admission from the firm side: GCs have said they don't necessarily expect bills to come down. What clients are loudly asking for is the conversation: proactively asking about their budget pressure, what value means on this matter, what the business outcome is. Almost none report getting that outreach.

Open that dialogue before a competitive process forces it, and you're the firm with the deepest relationship three years from now.

Value-add

The things that win cost almost nothing

Asked for the best thing a firm did in the last year that had nothing to do with winning work, the answers were strikingly small: picked up the phone to ask how we're doing. Recognized an associate who worked the weekend. Ran a seminar for the business. Helped deliver on our social-value goals. One client still cites, sixteen years later, a partner who sent a free two-page analysis before being hired, and still has the work.

From the room"Don't bill me for the call. Just pick up the phone and ask how we did. That's the whole thing."

Market shape

A bifurcation is coming, and the middle of the market feels it most

The buyers see the market splitting. The top tier has the capital to invest, lead, and price premium expertise for the work that gets charged to the deal. From there down, clients foresee a reckoning and consolidation, with firms combining to take cost out of the back office, alongside new AI-native entrants and a far larger role for alternative providers.

The clients also want the firms to win: those who partner on technology and shared wins will be in great shape; those fixed only on realization and rate increases will have a harder time.

The pitch

Do your homework. The stock pitch is the fastest way to lose

The most overused phrases the buyers are tired of: "client driven," and being "thrilled." Seven former judges and seventeen offices in a pitch deck don't move them. What does: evidence you've taken the time to understand their business and tailored the recommendation to it, the same standard you'd hold a candidate to in an interview.

And when you lose, ask why. Being told you lost because someone else delivered the value is the most educational feedback a firm can get.

Shared responsibility

This is one profession, and the relationship is going transactional

The most elevated thread of the day reframed the whole debate: in-house and firm lawyers are one community providing legal services to the world, and the language of "relationship" and "partnership" can't survive a dynamic where one side has to win. The risk isn't AI or pricing on its own. It's letting optimization quietly erode the trust the relationship was built on.

For in-house: own the narrative with your CFO before a Big Four pitch defines it for you. For firms: partner on the economics before a competitive process does the talking. Both sides agreed the answer is the same: a real conversation, in the same room, starting now.

The alignment gap

The disconnect isn't the lawyer-to-lawyer relationship. It's everything around it

Where the room aligned: trust, responsiveness, deep knowledge of the business, and a shared appetite for fixed fees that reward partnership. Where it diverged: procurement's role, what "better-defined expectations" actually means, who initiates creative pricing, and how AI savings get shared. The lawyers in the relationship are often aligned; the systems around them are not.

Closing that gap is a behavior change on both sides, not an agreement in principle, but a structured way to scope, price, and govern the work together.

Live Polling Data

Results from two live exercises run on the day: a nine-question AI maturity assessment completed by ~58 participants, and a table exercise with 12 tables designing an AI operating model for a real matter. Data is directional; firm-side samples are smaller.

The room, in one line

Everyone is looking over their shoulder to see where everyone else is on the AI journey.

72%/28%

Responses reflect 72% in-house and 28% firm — a mixed-room sample

~58

Responses per AI maturity question
9 questions, avg 52 responses

83%

Say AI governance should be
joint client–firm decision

★ The headline finding · Q9

AI operating posture: where does the room sit?

Using an assessment designed by Debevoise, attendees self-reported their archetype from five defined postures, ranging from Explorer to Scaled Operator. The market is concentrated in the middle; most are experimenting, not yet integrating.

Explorer

Aware and interested; no production use or formal program

54%
Controlled Experimenter
Piloting with guardrails; not yet at scale

33%
Workflow Integrator
Live in defined workflows; counterpart collaboration undeveloped

10%

Operating Model Builder

Redesigning scoping, staffing, pricing, delivery with AI

Scaled Operator

AI standard across delivery; joint client–firm models; measured outcomes

The market is concentrated between Controlled Experimenter and Workflow Integrator (87% of the room). No one in the room has reached Scaled Operator. The gap between where the market is and where it needs to be is the entire content of this day.

AI Maturity Assessment · 58 participants

Where the room sits on AI

Nine questions mapping AI maturity across workflows, governance, collaboration, economics, and measurement. Scale of 1–5 from exploration to full integration. Completed before sessions began.

Q1 · 58 responses

Where AI sits in your legal workflows today

1 — No approved AI use cases in legal work

2 — Individual experimentation; nothing repeatable

31%

3 — Defined use cases in 2–4 workflows with some governance

48%

4 — Embedded across workflows; most of team uses weekly

16%

5 — Standard part of delivery: integrated into staffing, scoping, pricing

48% have defined AI use cases with some governance, but only 2% have reached full integration into delivery, staffing, and pricing.

Q2 · 55 responses

When a new matter arrives that could benefit from AI, what actually happens?

1 — Nothing: AI isn't part of intake or scoping

18%

2 — Suggested informally; no structured path to approve or deploy

60%

3 — Defined assessment process, but often lands on "not yet"

4 — Pre-approved uses deploy at kickoff for routine matters

15%

5 — AI-assisted is the default; opting out needs justification

60% handle AI informally with no structured deployment path. No one has reached the point where AI-assisted is the default and opting out requires justification.

Q3 · 54 responses

If a lawyer wanted to use AI on confidential client data tomorrow. What would happen?

1 — No clear guidance on whether it's permitted

2 — Policy says "ask someone": answer takes weeks

15%

3 — Clear data-tool rules exist; edge cases need judgment

50%

4 — Risk-based governance with fast intake and pre-approved paths

11%

5 — Governance embedded in tools: boundaries enforced at point of use

19%

50% have clear rules but still rely on human judgment for edge cases. Notably, 19% have governance embedded in the tools themselves.

Q4 · 53 responses

How does your organization handle AI quality and verification?

1 — No formal approach: individual lawyers decide

2 — General "review AI output" guidance; no workflow protocols

75%

3 — Verification steps exist for priority uses; compliance uneven

11%

4 — Structured QA across most production uses, with clear ownership

5 — Verification built into workflow with checkpoints and metrics

75% rely on generic "review the output" guidance with no workflow-specific protocols. Only 4% have built verification into the workflow itself.

Q5 · 52 responses

Which best describes the AI conversation between your organization and primary outside counsel or client?

1 — We haven't discussed it

10%

2 — Mentioned in pitches or RFPs, but no operational detail

27%

3 — Discussed tools or workflows on a matter; no joint approach designed

56%

4 — Agreed protocols on specific matters: data, verification, pricing

5 — Shared AI working model: co-designed workflows, platforms, pricing

56% have discussed AI on a matter but have no joint approach. Only 2% have a fully co-designed shared working model.

Q6 · 48 responses

What would your counterpart need to tell you before you'd be comfortable with AI on shared work?

1 — Haven't thought about what I'd need to know

2 — Want to know they're using it, but unsure what else to ask

3 — Want specifics: tools, data inputs, human review, confidentiality

21%

4 — All of the above plus verification ownership, error handling, and pricing

67%

5 — Already have a working framework: question is how to extend it

67% want the full picture: tools, data inputs, human review, verification ownership, error handling, and pricing impact, before approving AI use.

Q7 · 48 responses

If AI meaningfully reduced time on a repeatable workflow, how would the economics be handled?

1 — We haven't addressed this

13%

2 — Efficiency flows entirely to one side: lower fees expected

15%

3 — Would discuss it, but no framework: negotiated ad hoc

29%

4 — Fixed fee, value-based, or shared savings models reflect AI efficiency

33%

5 — Pricing reflects AI delivery as joint design: both sides share value

10%

Only 10% have a genuinely joint model where both sides share the value. 33% are moving toward fixed fee or shared savings; the rest are ad hoc or unresolved.

Q8 · 42 responses

How does your organization currently measure whether AI is working in legal?

1 — We don't measure it

14%

2 — Anecdotal: "people say it's helpful"

31%

3 — Basic usage or time-saved metrics for some use cases

43%

4 — Defined KPIs (usage, quality, turnaround, cost) reviewed by leadership

10%

5 — Measured impact informs investment, staffing, scaling, and pricing

45% rely on anecdote or have no measurement at all. Only 2% have closed the loop so that measured AI impact actually informs investment, staffing, and pricing decisions.

AI Operating Model Workshop · 12 responses · Table vote

Who should own the final AI governance decision?

83%

say AI governance should be a joint client–firm decision

Joint Approval 83%

Client only 17%

Firm only 0%

Not one participant in the workshop said AI governance belongs solely with the firm. The mandate for a shared operating model is clear; what's missing is the structure to deliver it.

Unlock the full story