He Lost $50M Over One Sentence — Then Built This Law Firm

Collin McKee

Welcome back to the Endeavors AI Podcast. I'm Collin McKee, founder of Endeavors AI — my chance to talk to founders and operators about what AI actually looks like, minus the hype. Today my guest is Michael Woolf of Boon Legal, a virtual flat-fee law firm built for entrepreneurs. Before law, Michael was a tech entrepreneur in the game world, so he's been working with AI a lot longer than most lawyers. Michael, welcome.

Michael Woolf

Thank you so much — so excited to be here. I appreciate you having me.

Collin McKee

I want to start with the path, because it's not the usual one — you came into law sideways. Walk me through how you got from the game world into the firm.

Michael Woolf

It was obviously not a direct, normal course. I spent about twenty-plus years in the game industry, working on everything from art to design to directing products and getting into the executive suite, partnering in companies. Throughout that time — the game industry uses a lot of forward tech, so things like AI, high-end graphics, what would become UX/UI design all rolled out of that industry — we were always thinking, "What's the next thing to conquer? How do we make this a more realistic simulation? How do we get it playing with the player instead of just being an antagonist?" We experimented with all sorts of things, and we didn't label them — certainly we talked about AI as AI, but it was always "what serves the product." If that's the right thing to serve the product, that's where we'd go. I took a little of that with me into the legal world.

Collin McKee

You mentioned a deal that changed how you saw all of this. Tell us about it.

Michael Woolf

If you spend any time in business, the failures and setbacks tend to linger more — but they also teach more. I was working with a company on a piece of our own IP. In the game industry at the time, you sold the idea to a publisher/distributor, and more often than not they'd back your development along with the distribution deal. Those deals generally came with a set ROI. For example, it might be about $1.5M to press the discs and put them on the shelf, and a half-million for our development costs — so they're looking at $2M, and they get 100% of revenues toward that at the beginning, then a 4X ROI: the next $8M at 80% to them, 20% to the developer; once they reach the set ROI, it flips, so the developer gets 80% and the publisher 20%. Those games could be a fairly good payday in royalties for teams bringing their own IP. In this case, we did a half-million-dollar deal that ended up making somewhere between $53 and $56 million, depending on the numbers. And instead of getting what should have been upwards of $45 to $50 million, our company got zero in royalties. On reflection, it came down to essentially a single sentence missing from that contract. What stopped me further was that we had an attorney on that project helping negotiate the deal. Why did we miss out when we did everything "right"? The only answer I can come up with is that he was a well-regarded entertainment attorney — but in film and television, not the game industry. Those subtle, nuanced differences made all the difference. About fourteen months later, I was at a direct competitor in the same product line; we delivered two or three more products, and my two partners offered me a partnership. I said, "I'm grateful, thank you — and there's no way I'm doing this unless one of the three of us gets some legal education, because we can't let the 70 families that rely on us down because of our ignorance." That got the ball rolling. You remember those things more easily.

Collin McKee

So you put your business hat back on and asked, "What kind of firm would I want to walk into?" In building Boon Legal, what did you set out to kill in the traditional model?

Michael Woolf

Between law school and Boon Legal I spent some time with a giant law firm — big law, the 30th largest in the country at the time. The people in big law are fantastic; I have great friends there. But the structures taught over the last 150 years meant that all of my frustration — what I crassly called the "how much is this call going to cost?" moment, or "how much for three red lines on this document?" — those moments were frustrating as a consumer, and even more frustrating as a provider, because of the way the structure was set up. It's nobody's fault, except maybe a lack of creativity about how to change things or a lack of business knowledge. I stepped in saying my biggest frustration on both sides of that ledger is that communication is just horrible. So the first thing was: how do we fix communication? How do we make picking up the phone not tie your stomach in knots over cost? How do we let the entrepreneur and the business have someone learn the nuances of that business — what will take it forward or pull it back, not because of something in that vertical but a particularity in that business? It's very hard to do that when you're charging six minutes at a time — people don't want to pay for you to learn the nuances, but how do you protect somebody if you don't? That was step one. Step two was getting rid of the hourly-billing model. It's an absurdity for the provider — it works against both financial and time efficiency — and for the business owner, the speed of business and the speed of law aren't the same. So how do we front-load the businesses' priorities instead of where we sit? I just asked, "What would I want? I want to know what it's going to cost, when it'll get done, and whether someone will take the time to really learn about me." None of the traditional model felt right.

Collin McKee

That leads into your fee model — flat fee plus subscription. Explain how it changes the relationship with the client.

Michael Woolf

Start with the flat fee. Someone comes in: "I need to get XYZ done, can you help?" We explore it, decide we can, and here's what it'll cost. They can accept, negotiate, or reject — but once they accept, we don't have the issue a lot of folks struggle with: the bickering over an invoice. I saw this in software development too. We're now in our ninth year, and it's never come up once, because people won't argue with what they already agreed to on the front end, as long as it doesn't change and everyone thinks it's fair. And just because we call it a flat fee doesn't mean it doesn't get the love other work gets — if a contract needs to be finessed and nuanced, that's expected and built into the model. The recurring monthly plans aren't a requirement of any client, and there's no six- or twelve-month commitment — it's month to month. They provide ongoing service with enough parts for each growth stage of the business that clients can get 70, 80, or more percent of their legal covered under one budgetable expense. What clients on those plans tell us isn't even about using all the services every month — it's that they know the cost, they know they can pick up the phone anytime, and we'll never charge them to have a conversation. A friend of mine who runs a security company had clients saying, "We've been with you two or three years and don't see the need for all this security — nothing's going on." And he said, "Yeah — that's because it's working." Our clients appreciate that our conversations go from "what's the mess of the day" to "what's the strategy for the next three to five years" pretty quickly.

Collin McKee

I love that story — "nothing's breaking, so let's get rid of it." No, it's doing exactly what it's meant to do.

Michael Woolf

It's a tough one, because if there's a deterrent — an armed guard, or protecting your contracts — and bad actors aren't behaving badly because of it, it's hard to prove the value of what doesn't exist.

Collin McKee

Let's pivot to AI, since that's the reason for the show. You've worked with it a long time in gaming, well before ChatGPT. Where does AI live within Boon Legal?

Michael Woolf

AI is a pencil. That's the simple answer — it's a new pencil. It's what you make of it, like the printing press or the typewriter or the word processor. Everything can be abused if we choose to. I'm not one who has an idealistic view of AI — it has an extraordinary capability of causing great damage, but an equally great possibility of tremendous change for the better. With any tool, when we find we're doing something repeatedly, we ask two questions. First: is this brain work or finger work? If it's brain work, we work on it; if it's finger work, we build a process to take care of it. A good example: certain contracts we work on regularly have parts that are moldable around a particular use case. Take a master services agreement — they're functionally the same in what they do, but not the same in the application they're doing it for. So we let the AI handle putting together pieces while we focus on which pieces and how to finesse them afterward. We use it as a development-along-the-way tool, not a final-product tool. A lot of folks try to use AI to produce an end product, and that's premature in many situations, because what it really excels at is getting the 80% done that takes the most time — the old 80/20 rule — so you spend 20% of the work and 80% of your time making it the best for that particular client.

Collin McKee

You said something in our first conversation: it's not about the output, it's the instructions. Talk more about that — it's pivotal.

Michael Woolf

We often try to get to an output stage — "I'll hit this button and Claude or ChatGPT will go do this thing for me." There's a lot it can do and learn, and there are great people teaching how to think about the basics. But we look at it as a new pencil in the hand of a new person. Think about a law student who graduates and passes the bar — at that point they don't know how to practice law. They know how to pass a test that says they're allowed to, but not the practice part. We expect basic knowledge but a lack of context and experience to apply it the way someone with five, ten, fifteen, twenty years would. I think about AI the same way. Someone said AI today is like a five-year-old who happens to know nuclear physics — it knows academically how to do quite a bit because it looks it up, but the context, the purpose, the foundational frameworks behind things, it doesn't know. So we focus it on tasks that are either extremely well documented as to how you want things to turn out — like a product-development document for an app, where you lay out all the details — or you go the other way: "here's what I'm looking to do; figure out how you want to do it." Either way, don't expect the right answer. What you're trying to get to is the right process, because it's all about making good decisions that lead to good results. If you teach it why you're doing something — or ask how it would think about doing it — and it comes back with "I might do this, this, and this," you say "great, let's see what that's like, but tweak this one thing first." That's no different from working with a less-experienced person. I think the trick people miss is looking for a magic bullet or magic wand, and there is none.

Collin McKee

I've heard it equated to a new employee so many times, and it's so true. Give it the right instructions and guidance; if you expect perfection, your expectations are wrong.

Michael Woolf

Years ago, when I first became a manager, the owner of the company I worked for said, "The only thing I'll tell you about management — because you don't want to do what I do — is: you get to tell people what to do, how to do it, or when you want it. You don't get all three, and you shouldn't really do more than one, though sometimes you have to." There's great wisdom in that, and it applies to AI.

Collin McKee

For the listening audience — a lot are curious about AI, a lot are cautious, and they need to be. For someone who knows it matters but is nervous about wading in, where should they start, and what should they avoid?

Michael Woolf

Let's do the avoid first — that's easy: avoid sharing your data. It's not that we don't want to contribute to how it learns — plenty of people are doing that. Two reasons: once it's out there, it's out there, in whatever form. I'm told you could pinpoint any document or piece of information within the tools — I haven't tried it, it's not interesting to me, but I've heard that. They used to say about email, "don't put in writing what you don't want published on the front page of the newspaper" — same here. Also avoid the expectation that it'll be perfect. What you want is to develop a process and figure out which steps you need to do and which you can hand off to anybody else — whether that "anybody" is an AI or a person — without harming the end result. Those are the things to look at for the AI to tackle. To start: use the AI as a partner to have a conversation with. I write articles, and I've started using AI as a sparring partner. Over time I've taught it, "Remember I like to do this, I like things done this way." I'll feed it a bunch of things I've written and ask it to analyze how I write in different contexts and come up with a guideline for that style. I have a writing style for thought-leadership articles that's vastly different from my email style, which is different again from how I write a contract. Once it's learned those styles, it's simple to say, "Here's an email thread — in this voice, draft a first reply." Really simple stuff, but it's the 80% of the work that takes 20% of the time. We spend so much time on the first draft — just give me the first draft so I can rewrite it. In the words of Hemingway, "the first draft of everything is shit," so don't worry about it, just get it on paper. If a tool helps you get there in seconds, you can massage it. That's where you start to teach it what you do, and once it can do that repeatedly, you turn it into a custom GPT everyone can use. Just take it in steps — it's not much different from a new employee: "What can you show me today? Great, now let's work on the next thing." And it doesn't require onboarding — it's a mentorship more than an instruction.

Collin McKee

What I'm hearing is: take the time to teach it, and in that process you'll learn yourself what it has to give back to you. We're a culture that wants the quick answer, the easy button — but with this force multiplier, you really do need to take the time and be careful, and teach it what you want it to do. Great advice. Michael, where should people go to learn more about Boon Legal and reach out?

Michael Woolf

Thank you for asking. We're at boonlegal.com, and you can reach us there or email hello@boonlegal.com. We'll be glad to say hi and see if we can help.

Collin McKee

We'll put that in the show notes. That's Michael Woolf, founder of Boon Legal — find the firm at boonlegal.com. If you got something out of this, please follow the show. And if you're trying to figure out where AI fits in your business without going off the rails, that's what we do here at Endeavors AI — find me at endeavorsai.com. Thanks for listening, and we'll see you on the next one.