There are two versions of the AI conversation at every law firm I know.
The first version happens at partner meetings, in firm-wide emails, in carefully drafted memos from the technology committee. It says things like we are committed to thoughtful adoption of AI tools, with a focus on maintaining quality and protecting client confidentiality. It is balanced. It is professional. It is, in almost every respect, a non-statement.
The second version happens in elevators, on phone calls between partners who trust each other, in private conversations after the official meeting has ended. It is much more interesting. It includes things like Carolyn doesn’t realize her associates are doing all her research with Claude, and the partnership is going to fight over the billing model in three years and nobody wants to start that fight today, and I think we hire two fewer associates this cycle but I can’t say that out loud yet.
The gap between these two conversations—official and private—is, in my view, the single most expensive thing the legal industry is currently not addressing. It is also the conversation I have been writing this site to surface, because I do not think any of my colleagues will say it under their own names.
Why the official version exists
I want to be fair to the official version. It exists for understandable reasons.
Law firms are partnerships. A partnership is a relationship among peers, each of whom has built their career on a particular conception of what good legal work looks like. To say openly the model that produced your career is breaking is to say something deeply uncomfortable to a colleague who is also a partner, also a friend, and also entitled to vote on your compensation next year.
So we don’t say it. We say diplomatic versions of it. We commit to thoughtful adoption. We promise to maintain quality. We schedule trainings. The trainings are attended mostly by people who already understand the technology, performed mostly for the benefit of people who don’t and won’t.
This is not a moral failing. It is the social cost of being in a partnership of equals during a period when the equality is being challenged by external forces. The partner whose practice is most threatened by AI gets the most professional courtesy from the partners whose practices are not. We extend each other the dignity of pretending the disruption is gentler than it is.
I understand this. I participate in it. I would not survive at my firm if I did not.
Why the private version is more accurate
But the private version exists too, and it is almost entirely composed of the things the official version omits.
In private, partners discuss which of their colleagues are not adapting. They discuss which practice groups are quietly being staffed below historical headcount. They discuss whether to recruit out of firms whose AI policies have produced a class of associates with weaker fundamental skills. They discuss the mathematics of associate-to-partner ratios in a world where AI does what associates used to do.
These conversations are not malicious. They are the conversations that any group of professionals would have when their industry is changing under them. They are also the conversations where the actual strategic thinking happens.
The official version is what we say to maintain the partnership. The private version is what we say when we’re trying to figure out what to do.
What the firm tells the associates is the official version. What the firm tells the clients is the official version. What the firm tells itself, in its own internal strategic planning, is closer to the private version. But even the strategic planning gets softened, because the same dynamics that produce the official version at partner meetings produce the same softening in the planning documents. We tell ourselves slightly truer things than we tell our associates, but not by as much as we should.
The cost of the gap
The cost of the gap shows up in three places.
First, in associate development. Associates take their cues from what the firm tells them officially. The firm tells them officially that AI is a tool to be adopted thoughtfully. It does not tell them that the partnership privately expects associates to need to be twice as good as previous generations to justify the same career trajectory, because the work they used to do is now substantially automated. So associates calibrate to the official message. They use AI thoughtfully. They produce the same volume of output as their predecessors. They do not realize that the bar has moved without anyone announcing it.
The associates who figure this out, usually around year four, are the ones who pivot. The ones who don’t continue billing the old number of hours, producing the old quality of work, and are quietly not promoted. Nobody tells them why.
Second, in client expectations. Clients hear the official version too. They hear that the firm is committed to thoughtful AI adoption. They do not hear that the firm has internal estimates of what fraction of associate work is now produced with AI assistance. They do not hear that the partners themselves are quietly using AI to draft their own initial responses to client questions. The client thinks they are paying for human work, when in fact they are paying for human-supervised AI work. The supervision is real, the bill is real, but the underlying production is no longer what it was.
This is sustainable as long as nobody points it out. As soon as a major client points it out—and one will, eventually, with great public noise—the price renegotiation will be brutal. The firms that have prepared for this will survive it. The firms that have continued telling themselves the official story will not.
Third, in the professional ethics conversation. The bar associations are issuing increasingly detailed guidance about AI in legal practice. The guidance is well-intentioned. It is also calibrated to the official version of what’s happening, not the private version. It addresses lawyers who are using AI thoughtfully, with appropriate verification, in well-understood ways. It does not address lawyers whose entire workflow has reorganized around AI in ways that the official guidance does not contemplate. The gap between what the rules cover and what the practice actually looks like is widening every quarter.
When something goes wrong—a hallucinated citation, a confidentiality breach, a malpractice claim grounded in AI-generated work product—the bar will hold lawyers accountable to the rules, not to the realities. The lawyers caught in that gap will discover, in the worst possible way, the cost of having operated according to the official conversation while the real practice was happening privately.
What I tell people privately
When colleagues ask me, in elevators or in private calls, what I think is actually happening, I tell them something close to the following.
The legal profession is in the middle of a transition that is much larger than the official conversation acknowledges. The pace is uneven across practice areas. The pressure is unevenly distributed across age cohorts and seniority levels. The official message is wrong about the speed and the magnitude, but the official message will not be updated until enough firms experience the consequences that the message becomes untenable.
In the meantime, the lawyers who do best will be the ones who calibrate to the private conversation rather than the official one. That means: train yourself harder than the firm asks. Build skills that AI cannot replicate, not skills that AI augments. Notice which of your colleagues are quietly disengaging from substantive work and ask yourself whether you are in the same position. Have honest conversations about pricing with your major clients before they have those conversations with you.
This is, frankly, hard advice to take. Most lawyers I know prefer to operate inside the official version, because the official version is comfortable and predictable and lets them continue doing what they have always done. The official version will continue to be the comfortable choice for several more years. The lawyers who choose comfort will pay a price in the back half of their careers that they will not be able to diagnose at the time they’re paying it.
Why I write this site
This is the gap I started this site to address, in a small way.
I cannot have the private conversation under my real name, for the same reasons every other partner in my position can’t. The professional cost of public honesty is too high. The official version has to be maintained, in public, by the people inside the system.
But the private version exists, and it is more accurate, and the people who are calibrating their careers to the official version are getting bad advice. So I write the private version under a pen name. I know I am not the only senior lawyer thinking these things. I am, as far as I can tell, one of relatively few willing to write them down.
If you are a partner reading this and recognizing the two-conversations dynamic in your own firm, I would be curious to hear how you are handling it. If you are a younger lawyer trying to figure out which conversation to listen to, my advice is the private one—as much of it as you can get access to. The official version will not steer you wrong about etiquette. It will steer you wrong about strategy.
The bill for that error gets paid in the second half of your career. By then, the partners who told you the official version will be retired.
The most rewarding part of writing this site has been the private emails I receive from lawyers who recognize themselves in these articles. If you have a thought you cannot say at your own firm, send it to [email protected]. I won’t publish anything without permission, but I will read everything.