How to Choose an AI Automation Agency: 7 Questions to Ask

AI and automation are having a moment, which means a lot of agencies appeared overnight. Some are excellent; many will sell you a slick demo and leave you with a half-working system you can't maintain. The good news: a handful of pointed questions separate the two fast. Here's what to ask — and what the answers should sound like.

1. "Can you show me something running in production?"

Anyone can demo. Ask for a real client, a real problem, and a real outcome — a system that's live today. A good partner will happily point you at working examples (here are ours). Be wary of portfolios that are all mock-ups and "concepts."

2. "Who owns the code, data and configs when we're done?"

The answer must be you. You should be able to maintain and extend everything independently. If an agency keeps your system hostage on their platform or won't hand over the code and prompts, that's a dependency trap — walk away.

3. "How do you decide what to build first?"

The right answer starts with your business, not their favourite tool. A good partner asks detailed questions about your processes, data and costs, then recommends the highest-ROI place to start — often a small pilot. (It's exactly why we begin every engagement with an audit, and why we wrote a guide on the processes worth automating first.) If they're quoting a big build before understanding your business, they're setting you up for an expensive rebuild.

4. "What's the full pricing model — and the total cost?"

Look for transparent, fixed-scope pricing you can understand up front. Watch for per-seat licensing that balloons as you grow, and "it depends" that never resolves into a number. You shouldn't have to guess what you'll pay. (Ours is on the pricing page.)

5. "What happens after launch?"

Software isn't "done" at launch — apps change, needs evolve, things break. Ask for a written support plan: a defined response window, what triggers escalation, and what's included versus billed separately. A partner who builds and vanishes is worse than no partner.

6. "What would tell me it's working — or broken — in six months?"

A serious agency can describe exactly what you'd see: the metrics on a dashboard, and the signals that flag a problem before a customer complains. If they can't answer concretely, they haven't thought past the demo.

7. "Tell me about a project that didn't go to plan."

Everyone honest has one. The answer reveals whether they learn and adapt — or whether they'll bluff their way through your project too. Beware anyone whose track record is flawless; it usually means they're hiding something or haven't done much.

5 red flags that predict a failed project

• Vague "10× efficiency" promises with no link to your actual process, costs or error rates.
• Demo-only proof — nothing actually running in production.
• Won't let you own your code, prompts, configs or data.
• Quotes before asking questions about your business, data and users.
• No written support plan — just "we'll be around."

How to weigh it up

You don't need a partner who aces every question theatrically — you need one who's specific, honest, and happy for you to own the result. Real proof beats a polished pitch. Transparency beats "trust us." And a partner who'll tell you when not to build something is worth far more than one who says yes to everything. That's the bar — for us and anyone else you're considering.

The bottom line

Hiring an AI or automation partner isn't about who has the flashiest demo — it's about who can ship something real, hand you the keys, and still be there in six months. Ask the seven questions, watch for the five red flags, and you'll filter out most of the risk before you ever sign.