By: Andrea Flanagan

The alternative version of this blog post, if written for a lawyer audience, would be titled something like ‘how to not lose your mind every time a businessperson tries to pass off AI output as equivalent to your degree, experience, and expertise.’ Alex Su or Matt Margolis have probably already made hilarious LinkedIn posts poking fun at this new headache for lawyers. 

Instead, this post is for all y’all folks who work for the lean tech companies that are built on the premise that speed is survival. I’ve worked for those companies and as a co-founder of an ag-tech startup, I get it. I want to move fast, keep costs low, and get to the next milestone before running out of runway. Headcount is lean by design, outside counsel feels like a luxury, and AI legal tools have gotten good enough that the gap seems bridgeable. For a lot of everyday tasks, it is. I use AI tools in my own work to refine content, teach myself unfamiliar territory quickly, and to get oriented before making decisions. When we were building a feature to help farmers navigate federal grant applications, AI helped me get up to speed on grant compliance faster than I could have on my own. 

But as a general counsel, I know fast isn’t always better, especially without limits. AI can help your team educate itself. It can help you generate a first draft of a simple NDA. What it can’t do is reliably protect you when the stakes are high, when you’re negotiating against a sophisticated counterparty, or when the decision you’re making today will shape your company’s legal posture for years. I’ve seen what happens when speed outpaces diligence: what should be hiccups become halts and overlooked clauses that resurface at the worst possible moment. 

Learning to see that line and being honest with yourself about which side of it you’re on is one of the most important skills you can develop. My husband, a twenty-year Army infantry vet and engineering VP, has a phrase for it: slow is smooth, smooth is fast. It’s a military principle, often repeated without conviction in corporate echo chambers, but applies all the same. Sometimes the fastest path forward is the one where you slow down first.

The “good enough” trap

AI tools are genuinely useful for a certain category of legal-adjacent work. They can break down a regulatory framework in minutes and point you toward the primary sources worth reading. Quality varies. Some tools cite sources and flag jurisdictional uncertainty; others answer with equal confidence whether they’re on solid ground or not.  The tools are good enough to get you most of the way there and offer, at face value, a ton of value for a lean team trying to move quickly. 

But the distinction between useful guidance and actual legal advice isn’t always obvious, especially if you don’t work in law. A lawyer reading AI-generated legal guidance knows what to verify, what to push back on, and what should trigger a deeper look. A founder running lean without legal training doesn’t have that framework and often doesn’t know they need it. The output looks authoritative whether it is or not. Not to mention the tendency for all models to be agreeable. That’s the trap.

Knowing when to slow down

Given enough context and clear prompting, generative AI can customize a document to your company’s preferences, flag clauses that don’t fit your situation, and turn a first draft into something polished. For internal documents and early-stage templates, it works.

What AI cannot do is sit across the table from another human being (recent reports of a certain social media tycoon’s effort to create a digital avatar of themselves is not lost on me). In the moments that matter most to a lean tech company, there is typically a person on the other side with their own interests and their own read of the room – that is not the time to defer to AI. Those conversations require someone who can sense when to push, when to concede, and when something said casually in a meeting has legal weight that needs to be addressed. Those moments require a relationship, not just a document. This is where slow is smooth, smooth is fast earns its keep. 

The lean company that treats a fundraising negotiation or a co-founder agreement as something to get through quickly and cheaply is all but guaranteed to revisit that decision later. Down the road when fixing it costs far more than slowing down and getting it right would have.

Not every legal question requires outside counsel. Part of using AI tools wisely is developing an honest sense of when they’re sufficient and when they’re not. That discernment is more predictable and easier to apply than it might seem. AI can educate and outline. An internal team member with relevant context and experience can pressure-test. Counsel can then review and iterate on something already substantive rather than starting from scratch. That workflow doesn’t eliminate legal spend; however, it can meaningfully reduce it, which makes the conversation about bringing in counsel easier to have.

On that note: lawyers are not a necessary evil. They are assets, and the lean tech companies that treat them that way tend to fare better than those that avoid them until there’s no choice. Phil Knight surrounded himself with lawyers and accountants early, before Nike was Nike, and it’s hard to argue with the results. The goal is to be strategic about when and how you rely on lawyers and strategic patience is exactly what slow is smooth, smooth is fast means in practice.

The bottom line

AI is not going away, and neither is the resource pressure that makes lean tech companies reach for it. The tools are getting better, and legal service models are catching up. The gap between good legal advice and what early-stage companies can access is narrowing.

Use AI as a resource. Use it to educate yourself, to draft a starting point, to cut down on billable hours by showing up to outside counsel with something already substantive. Just don’t use it as a replacement for actual qualified legal advice when the stakes are high, when there’s a sophisticated party across the table, or when the decision you’re making today will be very hard to unwind tomorrow.

Slow is smooth. Smooth is fast.

Andrea Flanagan is a Certified Information Privacy Professional (CIPP/E) and currently serves as General Counsel at Yomali Labs Ltd.