You own an app you didn't entirely write. Something's off. And you're staring at the fork every AI-era founder hits weekly: do I ask the AI to fix this myself, or is this the moment to pay a professional?
Both answers are right — for different problems. The trick is telling which kind you're holding. Two questions do most of the work.
Question one: what's the blast radius?
If this change goes wrong, what does it touch? A typo on the pricing page has a blast radius of one page. A change to your checkout flow has a blast radius of your revenue. A database migration has a blast radius of everything.
Be honest here, because AI-built apps hide connections — the AI wired things together in ways you've never had to look at. When in doubt, assume the radius is bigger than it looks.
Question two: can you undo it?
Reversible means: you know the exact steps back to the working state, and you've confirmed they work. A deploy you can roll back is reversible. A database change usually isn't — not really — unless you have a backup you've actually restored before. "The AI can probably fix it again" is not a rollback plan. It's a second bet, placed while losing.
The quadrants
- Small radius, reversible: go for it. Copy changes, styling, layout, adding a FAQ section, tweaking email text. This is where AI-assisted self-service genuinely shines, and paying someone would be a waste. Have fun.
- Small radius, hard to undo: proceed, but slowly. Anything touching stored data in small ways. Make a backup first, change one thing, verify, then the next.
- Big radius, reversible: try it — off-hours, with the rollback steps written down before you start, and a test of the money paths right after. Dependency updates live here; we wrote a whole routine for them.
- Big radius, irreversible: stop. This is the quadrant where founders turn a bug into a catastrophe. Auth changes, payment logic, migrations, anything that deletes. Get help — not because you're not smart enough, but because the downside is unbounded and yours.
The cost math nobody actually does
Founders compare "free (me + AI)" against "$150 (a professional)" and choose free. But the real ledger reads differently. Count the evening you spent, the second evening when the fix broke something else, the customer who hit the bug meanwhile, and — the big invisible line — the risk you carried in that fourth quadrant.
Your building time is worth something too. Every hour spent nervously patching checkout is an hour not spent on the product. Somewhere around "real revenue flows through this app," the math flips for the risky quadrants, and it flips hard.
If you do hire help, brief them like this
Five lines make a hireable brief: what the app does, what's wrong (with a screenshot or error text), when it started, what you already tried, and what stack it's on — even roughly. "Built with Lovable, Supabase behind it, payments via Stripe" saves an hour of discovery, and you know that much even if you've never opened the code.
And one demand you should always make, from anyone, including us: ask what their rollback plan is before they touch anything. A professional will have an answer ready. If they don't — that's your answer about them.
The pattern worth noticing
If you find yourself in the danger quadrants monthly, the question isn't "who fixes this one" anymore — it's "why is nobody accountable for this app's operations?" That's a standing role, not a series of gigs. It's the role we take with Managed Operations, SLA and all. But whichever way you solve it: the two questions stay. Blast radius. Reversibility. Ask them every time, and you'll be right far more often than the founder who just types "fix it" and hopes.