Firm AI programs rarely end because the technology fails; they end when partners stop believing the program is worth its cost. I've watched that belief form and fade in steering meetings. The software can be working and improving every week, and the program can still lose its footing because the partnership never saw a result it recognized.
The stakes are rising. The 2026 Report on the State of the US Legal Market from the Thomson Reuters Institute and Georgetown Law found that the average firm grew technology spending 9.7% and knowledge management spending 10.5% in 2025, and coverage of the report described firms as making unprecedented bets that AI-enhanced capabilities will justify premium pricing. Bets that size get reviewed at partner meetings. Every dollar the program spends comes out of distributable profit, and partners know it.
Why partner patience runs out
Most partners aren't hostile to AI. Many of them use it every day. Their patience with a firm program runs out for three ordinary reasons.
The first is that the results are invisible to them. A program that spends six months on data pipelines, security reviews and integrations may be doing exactly the right work, and none of it shows up in a partner's day.
The second is that the program is described in technical terms. Deployment models, pipelines and benchmarks mean a great deal to the team building the system. A partner wants to know which clients it helps, which matters it touched, and what it changed about the work.
The third is that the economy moves. A program that looked affordable when a practice was busy looks different when deal flow slows or interest rates stay high. Discretionary spending is the first thing a firm reviews in a downturn, and a program with no visible client value is easy to pause.
Set a 90-day target a partner can see
The first milestone should be something a partner can use, or can watch a client use. In the programs I work on, the first 90 days aim at two things: a stable application in daily use by a real practice group, and a pipeline that brings the firm's documents in reliably. The first shows the program delivers. The second is the foundation everything after it depends on.
The target has to be concrete and dated. "Improve knowledge management" can't be checked at the end of a quarter. "A named practice group uses the tool on live matters by December 15" can.
Put it in front of clients early
The strongest argument for an AI program is a client who uses what the firm built and asks for more. Partners care about client relationships above almost everything else, and a client who is engaged with a firm-built tool changes how the partnership talks about the program.
Early client access doesn't require a finished product. It works best as a collaboration: the firm shows what it has, the client says what it needs, and the tool is shaped around the answer. Some firms charge a modest fee for that early access so both sides take it seriously. In my experience, partners care much less about what a pilot client pays than about whether the client is engaged.
Describe the program in business terms
Build one document that explains the program the way a partner would explain it to a client: what it does, which matters and clients it serves, what it has delivered so far, and what comes next. Keep it to a few pages. Use it at partner meetings, with the management committee and in front of clients. When the same document works in all three rooms, the program has a story partners can repeat without help.
Keep the technical detail in an appendix. Partners will ask about security and privilege, and the answers should be ready, but the body of the document is about clients and matters.
Use the steering meeting for decisions
Many AI steering meetings drift into status updates. Status belongs in a short weekly written note. The steering meeting is the one room where the program's sponsor, the firm's technology leadership and the delivery team can talk openly about what's blocked, what clients are asking for, and which decisions need a partner's support. When it runs that way, the sponsor walks into every partner meeting knowing what to say and what to ask for.
Report the measures partners trust
Partners trust measures they can connect to their own practice. These hold up well:
- Named adoption. Which practice groups and which partners use the program on live matters, by name.
- Client use. Which clients have seen or used what the firm built, and what they said about it.
- Reuse of prior work. How often lawyers found and used a position, clause or analysis the firm already had.
- Answers from the firm's own record. Questions answered from the firm's documents, with citations a partner can check.
- Risk handled. Conflicts flagged, access requests refused, and privilege questions resolved with an audit trail.
Thomson Reuters also reported that firms with a formal AI strategy are 3.9 times more likely to experience critical benefits from AI. A written strategy with dated targets and measures partners recognize is what a formal strategy looks like in practice.
Plan for a slower year
Every program needs an answer to the question a managing partner will eventually ask: what happens to this if the market turns? The programs that survive a slowdown are tied to revenue-producing practices and to named client relationships. When a program is part of how the firm serves an important client, pausing it becomes a client decision, and firms protect those. Getting the program into the hands of partners and clients quickly is the best protection a program has against a slower year.
The same logic applies to cost. A program whose cost rises with every lawyer who adopts it gives partners a reason to slow adoption at exactly the moment the firm needs more of it. Fixed, predictable costs make the program easier to defend when budgets tighten.
How Reframe approaches it
Reframe starts with one practice group and the firm's own documents, so the first results show up on matters partners already know. Every answer is cited to the document it came from, the system runs inside the firm's own cloud, and every finding arrives as a proposal a person approves. The license is fixed, so partners know the cost before the program starts, and it doesn't rise as more lawyers use it.
The work that keeps partners behind a program is simple to describe. Show them a dated result, put the program in front of clients, and explain it in the terms a partner would use with a client. Programs that do those three things early get the time they need to pay off.
Drafted with Reframe Editorial Agent, an AI employee at Reframe. 