Series A documents came in for a client last week. I got on a call with the founder and we walked through them: what to push on, what to concede, what needed to change. At a big firm, the next step is an associate spending hours rebuilding context, re-reading the email thread, pulling prior drafts, checking notes so nothing promised on the call gets lost. At most firms, that context lives in whoever happened to touch the matter, and reconstructing it is billable work.
I asked our system to do it instead. Review the email correspondence with the client. Review the notes from the call we just finished. Review the prior iterations of the documents. Then revise the drafts to reflect what the client and I decided, drawing on every financing the firm has papered before this one. The documents came back turned, grounded in the full history of the matter, in minutes.
That’s the firm’s side of the story. The founder’s side is simpler. They got their documents back the same day and a stronger position, with no bill for the hours of context reconstruction that usually pad a document turn. That’s what the memory is for. It puts senior judgment in front of a founder quickly, at a price we can name in advance.
We’re a small firm and we routinely sit across the table from firms with hundreds of lawyers. We hold our own because the firm itself has become an institution that remembers what it’s done. So every client works directly with a senior lawyer who has the whole firm’s experience behind them, instead of getting handed to a third-year learning on their deal.
This is about how we built this system, what it actually does, and where we’re still figuring it out.
The individual lawyer is the first chapter, not the whole book
If you’ve read Zack Shapiro on practicing law with Claude, you’ve seen the individual version of what we’ve built. One lawyer encodes his own judgment into reusable instructions, hands first-pass drafting and review to AI under his supervision, and produces the output of a much larger team. We do this too. We trained the system on how we review contracts and mark up a term sheet, so the first pass starts at our standard instead of a generic one.
Shapiro makes a point I think is correct, that the leverage is in the individual lawyer’s encoded judgment, not in some vendor’s idea of a firm template library. And he’s right about the harder limit too. His point is that you can feed a model the record of every deal you’ve ever done, but not the experience of having been in them. Michael Polanyi called that tacit knowledge, the stuff we know but cannot fully write down. It doesn’t transfer cleanly to a machine, and it may never.
So we didn’t bottle our judgment and just walk away; tacit knowledge doesn’t transfer that cleanly. What we built is a memory, and it goes after a problem the individual approach leaves untouched. When the record of what a firm has done lives in whoever happened to do it, the knowledge leaves when that person does. A vacation, a busy week, a departure, and it’s gone. That’s the oldest problem in professional services, and giving every lawyer their own AI does nothing to solve it.
A firm is two things, its memory and its positions
Plenty of legal-AI companies sell a version of what I’m about to describe: the firm brain, the platform that makes us smarter. But I’m not selling software. I built this inside a working firm and use it on real deals, which is a different thing from a demo.
Strip away the letterhead and a firm is two assets. The first is everything it has ever done. Every agreement it has drafted and every deal it has negotiated, down to the call someone made under pressure at eleven at night. The second is how it decides: the positions it takes on a given term, where it pushes and where it gives way, and what it treats as market in a sector this quarter. At almost every firm, both stay in human heads and scattered files. Neither is searchable.
We made them searchable. The system has read everything the firm produces: every document in our document management system, our emails, our meeting notes, our Slack. So when one of us has a question, we ask the firm instead of our own memory. What’s our standard position on a SAFE valuation cap? Have we seen this protective-provision language before, and how did we handle it? The answer comes back grounded in our own work, with the source attached. It only holds what we have actually done. It was thinner when we started and gets deeper with every matter we close.
The obvious question, and the line we drew
We know that if a system has read everything the firm has done, the first thing a careful founder or a general counsel is going to ask is, what about confidentiality?
The system runs on enterprise infrastructure that does not train on our data and does not retain it outside our control. Client information stays inside the firm’s confidentiality perimeter, full stop.
We treat anything that draws on prior client work as governed by the duties that have always applied, confidentiality and our obligations to former clients (Model Rules 1.6, 1.9, and 1.7, and the ABA's 2024 guidance on generative AI in Formal Opinion 512). The ABA is explicit that a boilerplate line in an engagement letter does not amount to informed consent for this. So our engagement letters say, in plain language, how prior-matter experience may inform later work, and we get actual consent.
The case I think about most is the temptation to tell a founder what is market in their sector this quarter. Lawyers have always carried that knowledge from doing the deals, and the system makes our own experience easier to recall. But there’s a line between drawing on our judgment and pulling a specific client’s confidential terms into someone else’s negotiation, and we do not cross it. Rule 1.9 is explicit that you cannot use a former client’s information to their disadvantage, so where a matter touches a client's competitor we wall it off, and where we cannot answer without exposing a confidence we say so and stop.
What’s live, and what we are still building
The memory and the positions are live and in daily use. But we have more still to build.
Think about what a firm like ours does over and over: qualify a new matter, form the company, paper a financing, review an inbound commercial agreement, run diligence on an acquisition, turn a board meeting into minutes, keep filings current, get the bill right. Each of those has a recognizable shape. A precipitating event, and somewhere in that routine work that follows, a lawyer makes the calls that really matter.
We’re turning each one into a supervised loop. The system notices the trigger, does the first pass against the firm’s memory and positions, and hands a lawyer a draft that already starts at our standard. A lawyer reviews it and signs off, or sends it back to be fixed. Nothing reaches a client that a licensed attorney has not reviewed and approved. And we’re trying to build each loop to learn so that every correction a lawyer makes improves the system for the next matter.
The system gives us a negotiation brief that shows how a specific opposing firm has actually moved across every deal we have done against them, where they open and where they will not budge; or a shortlist of the right investors for a founder about to raise, drawn from years of relationships and how those introductions actually played out. None of that’s possible without the memory underneath it, and each one gets better every time we close another matter.
What changes
Start with the honest version of the numbers. My own estimate is that the system makes me roughly a third better at my job. But the number understates the change. Without it, I would have to run fewer matters at once, and I would drop balls, because the context would live only in my head and my head cannot hold all of it. The point of a memory is that it does not have to.
The math on hiring changes. The work that used to justify the next associate (first-pass review, research, initial drafts, digging precedent out of old files) is increasingly done by our firm’s system, under supervision. To be clear, every document that leaves here is reviewed, revised, and signed by an attorney. The AI produces the first pass, and we produce the work product. So our next hire is about judgment and client relationships and the supervision this system needs, not another two thousand billable hours of production. The shape of the firm changes with the math. The traditional model is a pyramid: partners at the top, associate leverage underneath, profit flowing up from hours billed below. We are building a rectangle instead, senior lawyers made more productive by the system, with junior hires doing only what the technology cannot. We grow by getting smarter rather than just bigger.
This enables a new approach to pricing. When every hour of attorney time produces more, and routine advisory is significantly handled by the system, fixed-fee and subscription pricing become more realistic. For a growing company, the payoff is counsel you can call without watching the meter, which is what clients have wanted from lawyers for decades.
What stays human
AI isn’t practicing law. We are. It makes us faster and more thorough, and it leaves the decisions that matter to us. If you’ve spent fifteen years building that judgment, you have the exact asset this technology makes more valuable.
We drew one more line deliberately, and we wrote down which parts of the work will never become a loop. Matters like the hard conversation with a founder about a cofounder split or a layoff, or a negotiation that hinges on reading the room. Or the deal structure we haven’t seen until now. Or putting your name on a delicate call that could go either way. Those matters stay human, permanently and by design.
Why I’m writing this
I think the legal profession is about to split into firms that understand this and firms that do not, and I would rather model how to employ this technology prudently and creatively.
We’re looking for the people, clients and lawyers both, who find this exciting rather than threatening.
Most of the legal-AI conversation is about a better chatbot for one lawyer. The question I care about is whether a firm can become an institution that remembers what it has done and still leaves the important calls to the people whose names are on them. That’s the firm we’re building.