When Three Partners Disagreed on AI Strategy (And How We Resolved It)

A 14-attorney firm. Three equity partners. Six months of arguing about AI. They could not move.

Partner A wanted to ban all AI tool use until the bar's guidance was clearer. Partner B wanted to be aggressive, build internal tools, market AI capabilities to clients. Partner C wanted to "wait and see what other firms do."

Their AI strategy meeting in December went 90 minutes and ended with no decision. Their March meeting went 110 minutes and ended with no decision. Their May meeting was scheduled and the managing partner called me.

This is the framework we used. It got them to a unanimous vote in 90 minutes. The firm is now 8 months into a clean, governed AI rollout.

Why the disagreement was actually stuck

When I sat with each partner individually, the disagreement wasn't really about AI. It was about risk tolerance, brand identity, and which clients they wanted next.

Partner A had a major regulated client (a public company) and a near-miss earlier in their career with a privilege issue. AI registered to her as "another way to lose a license."

Partner B had a corporate practice in tech-adjacent industries. Her clients were already asking what the firm was doing with AI. For her, "no policy" was costing her business.

Partner C wanted to retire in 5-7 years. For him, AI was a question of "is the firm I'm selling worth more or less if it has invested in AI." He genuinely didn't know.

Three different fears. Three different time horizons. Three different views of the client base.

You can argue tactics for 90 minutes a quarter and never get to the underlying disagreement. So we didn't argue tactics.

The framework

I built a 2x2 matrix for the room.

X-axis: risk to the firm if we go faster (privilege, malpractice, ethics, reputation). Y-axis: risk to the firm if we go slower (client loss, talent loss, market position erosion).

Then I made each partner privately rate every proposed AI initiative on both axes. We did 14 initiatives. Things like "use Lexis+ AI for research," "use Claude for first-draft contract review," "publish AI capabilities on the website," "offer AI-augmented services as a separate billing line."

Then I plotted all three partners' ratings on the same matrix and we looked at it together.

The result was clarifying. On 9 of 14 initiatives, all three partners agreed within one rating point. They had been arguing about 5 specific things. Not 14.

The 9 they agreed on, we approved that day. Things like "everyone uses Lexis+ AI" and "associates can use Claude for research with disclosure to supervising partner" — unanimous.

The 5 contested ones we mapped to a decision framework. Each one got a specific question to answer.

The contested 5

Each contested initiative had a partner-specific concern that, once articulated, could be addressed.

1. **Publishing AI capabilities on the website.** Partner A was worried it would attract the wrong clients and expose the firm to scrutiny. We resolved it by publishing capabilities in a careful tone, framed as risk-managed adoption, on a deep page (not the homepage). A's concern was specifically about positioning, not about whether to mention AI. The specific positioning solved it.

2. **AI for first-draft contract review.** Partner B wanted to start immediately. Partner A wanted to wait for state bar guidance. We resolved it by piloting on the firm's own internal contracts (vendor agreements, etc.) for 90 days, then re-evaluating for client work. A learned the technology in a no-risk context. By the 90-day mark her concerns had moved from "ban" to "specific guardrails."

3. **Billing AI as a separate line item.** Partner B wanted to. Partner C thought it would confuse clients and damage the firm's pricing power. We resolved it by tabling for 12 months. B got the win that the firm would adopt the underlying capabilities. C got the win that they wouldn't restructure pricing during a transition.

4. **Hiring an AI implementation lead.** Partner C didn't want the headcount. Partner B thought it was the difference between adoption and abandonment. We resolved it by engaging me (an external consultant) for 6 months instead of hiring. C got the win that it wasn't a permanent FTE. B got the win that someone was actually responsible for the rollout.

5. **Marketing the firm as "AI-forward."** Partner A vetoed this entirely. Partner B accepted the veto in exchange for B getting to participate in industry panels and publish authored pieces about responsible AI adoption. The firm's brand stayed conservative. B got to build her personal brand in the space.

Why the matrix worked

The matrix worked because it forced each partner to defend a position with specifics. "I'm against AI" is hard to argue against. "I rate this specific initiative as 4 out of 5 risk-of-going-faster because of privilege concerns from my regulated client" is something we can talk about.

It also revealed that the partners actually agreed on most things. They were arguing about the loudest 5 initiatives and assuming they disagreed on everything. They didn't. Once they could see the agreement on 9 of 14, the room temperature dropped 40 degrees.

What I learned about partner-led firms

The technical problems in AI implementation are not the hard part. The partner alignment is the hard part.

Most partner groups have 2-3 partners with strong opinions about AI. Those partners are usually disagreeing not about AI but about other things (risk tolerance, brand, succession, client mix) that AI happens to surface. If you address the AI question without addressing the underlying disagreement, you'll be back in the same room in 3 months.

If you run an AI initiative inside a partner-led firm, the first deliverable is partner alignment, not a system. The system is week 8. The alignment is week 1.

What I'd tell another firm

Don't bring vendors to the partner discussion. They will pitch. You don't need a pitch. You need a framework that surfaces the actual disagreement.

The 2x2 matrix above is one framework. There are others. The key property is that it makes each partner state a specific concern about a specific initiative, on the record, in front of the other partners. That alone resolves 60% of the apparent disagreement.

If you can't get partner alignment, don't build anything. You'll spend the whole year explaining yourself to the dissenting partner. The cost is higher than the project's value.

If you can get partner alignment, the rest is mechanical. The mechanical part is what I do for a living. The alignment part is what you have to lead.