Started an AI-native law firm for Canadian founders and SMB. First month revenue tracking at low-to-mid five figures. (I will not promote)

Background: I am a Canadian lawyer with 25 years of contracting experience ranging from one of the biggest law firms in Canada, to 10 years as general counsel at one of BC’s biggest private companies. I was also CEO at an international aviation contracting business before retiring and deciding to build something new. A few months ago I started building an AI-native law firm. Flat fees, 48 hour turnaround, based loosely on the model Crosby and General Legal have been building in the States. My tech partner built the back end. AI does the first pass on every matter, drafts and redlines, working off our own playbooks and templates. Every draft is versioned and carries an audit trail. We run on commercial terms so client data isn’t used to train the models. I read, correct and sign everything before it goes out. Watching the back half of my career, I became convinced AI was going to significantly change how businesses work with legal, and I wanted to build a firm that reflected that from the start rather than adding AI later. Demand has been constant since we opened. We’re live and serving clients across SaaS agreements, contract negotiations and employment work. Our first month isn’t closed yet but it’s already tracking to low-to-mid five figures. Going through this volume of contracts in a short period has exposed some surprisingly consistent problems in startups that have handled a lot of their own legal work, either with templates or increasingly with AI. Most of the examples below are Canadian because that is where I practise. If you’re in the US, the specific law is obviously different, including from state to state. But the questions founders should be asking are much the same. 1. Your agreement may be based on law that doesn’t apply to you. I regularly see Canadian agreements with US at-will employment language, CCPA provisions, GDPR-heavy privacy language, or governing law and arbitration clauses that appear to have simply come with the template. AI can produce an extremely professional-looking agreement that is legally built for somewhere other than where your company actually operates. If you’re in the US, don’t take the Canadian examples as applying to you. The question is the same though: was this agreement actually built for your company and jurisdiction, or did the template or AI simply choose one for you? 2. You may not actually own the product you’re building. In Canada, if an independent contractor writes your code or designs your product, they generally own the copyright unless they signed a written assignment. Paying them does not transfer it. So the freelancer who built your MVP two years ago may still own part of the IP. You often don’t discover that until you raise, sell the company, or someone asks for chain of title in diligence. The US rules are different, but if you’re an American founder the practical question is still worth asking: can you actually prove the company owns everything its contractors built? 3. Your liability cap may not cap the liability you’re worried about. A liability cap can look reassuring until you realize indemnities, IP, confidentiality or data obligations have been carved out of it. So you read the cap, feel covered, and sign, while some of the largest potential exposures under the agreement sit outside it. This one isn’t particularly Canadian. Whatever jurisdiction you’re in, the useful question isn’t just “what is the cap?” It’s “what isn’t covered by the cap?” The recurring theme isn’t that AI is bad at legal work. I use it every day. It’s that an agreement can be polished, coherent and look completely professional while still being wrong for the company using it. If you’re a founder, I would take another look at your customer agreements, contractor agreements, IP assignments and privacy documents with that in mind. Use AI if that’s how you work, but tell it the jurisdiction and specifically ask it to identify provisions that assume a different one. These are much easier problems to fix before a financing, acquisition, dispute or major customer finds them for you. Happy to answer questions about any of this, or about what we’re learning building an AI-native law firm.