When to use AI tools yourself vs. hire help

“Should I just build it myself with AI, or hire someone?” It's the question I get most, and the answer people expect from someone who does this for a living is “hire us.” So let me be honest instead: for a lot of what you're trying to do right now, you don't need us. DIY-with-AI is genuinely good for some things, and I'd be doing you a disservice to pretend otherwise.

I've spent 22 years in startups — healthcare, security and privacy, the gig economy, commercial contracts — and I've watched the cost of trying an idea collapse to almost nothing. That's a good thing. The trick is knowing which side of the line you're on, because the same tools that are perfect early can quietly get you in trouble later.

It's perfectly fine to explore ideas with AI — experiencing the product you think you want, and refining ideas. AI is perfectly suited for that. I start my own ideas by creating prototypes I can experience, because before I validate an idea with others, I want to know if I think it's worth moving forward with. AI has made that incredibly easy.

There's something powerful about holding a rough version of your idea in your hands. You learn more in ten minutes of clicking through a thing you built than in ten pages of planning it. So if that's where you are — poking at a concept, deciding whether it's worth pursuing, getting a feel for it — go build. Cheaply, quickly, by yourself. This is exactly what the tools are for.

Here's a simple way to think about which side you're on:

• Green light, do it yourself: exploring an idea, prototyping to experience it, refining what you actually want, deciding if it's worth pursuing.
• Get help: real users relying on it, promises you've made, anything people will trust with their data or their money.
• Still unsure? If breaking it would let someone down, you've crossed the line.

Using pure AI to develop becomes a mistake once you start taking on real responsibilities — once you're making promises and delivering things users will expect, and worse, rely on. Then it's incumbent on you to ensure the application they're putting their trust in is developed properly.

That's the line. Not a particular feature, not a number of users — it's the moment someone other than you starts depending on the thing. The shift is from “I'm learning whether this is worth it” to “people are counting on me.” Those are two completely different jobs, and the second one carries weight the first one never did.

The moment someone other than you depends on it, you've crossed the line.

I hear stories all over social media about vibe-coded apps leaking data, leaking secrets, creating unexpected behaviors, and undermining people's confidence in software. I avoid those apps. At the end of the day, if you didn't put a lot of effort into your application, it shows — whether you're using AI or not. Craft, care, and attention to detail still matter, and reflect in your end product.

There's a deeper problem underneath the headlines, too. When you actually understand how your app is built, that understanding is what lets you trust it. If you skip that, you can't fully trust your own app — you're blind to its gaps. You won't know what it does in the situations you didn't think to test, and you won't find out until a user does.

I treat AI like a developer I'm coaching: useful, fast, worth the investment, but never the one in charge. It's leverage, never the hero. Somebody who understands the work still has to be accountable for it. When nobody is, that's exactly when the quiet failures creep in.

So here's the path I'd actually recommend, and it's not “hire help from day one.” Validate cheap with AI. Build the prototype, experience it, refine it, decide whether it deserves to exist. Most ideas don't survive that step, and you'll have spent next to nothing to find out. That's a win.

Then, the moment it earns real users and real promises, build it properly, with experience behind it. How do you know you've crossed over? Ask yourself one question: if this broke tomorrow, would it let someone down? If the answer is yes, you're no longer prototyping — you're running a product, and it deserves to be treated like one.

If you're still exploring, go enjoy it — genuinely, that part is yours to do. But when you cross that line, when people start relying on what you've built and it has to hold up, that's what we're for. We turn the idea you've validated into a real, launchable product you fully own.