Before most people hire me, they email the same set of questions. This page is the full list, answered honestly. If your question isn't here, ask it and I'll add it.
How much does Steve the AI Guy charge?
The first build is free. No deposit, no setup fee, no "waived if you sign." Actually free.
After it's live and working, the monthly retainer is typically $600–$1,500 AUD depending on complexity, with no lock-in contract. If the build doesn't move your numbers, you owe nothing. I've written in depth about the commercial logic in why I build the first one free.
What happens if the AI agent breaks?
The retainer covers monitoring and fixes. If anything goes down, I notice before you do — every live agent has uptime monitoring, and customer-facing outages get treated as urgent.
Typical response time is well under an hour during business hours. There's also always a fallback path built in, so a broken agent doesn't mean lost customers — calls route to you, forms land in your inbox, whatever the appropriate degrade-gracefully path is.
Am I locked into your tooling?
No. Everything I build for you is documented, yours, and portable. No proprietary platform, no exclusive contracts.
I use industry-standard tooling — OpenAI and Anthropic models, Retell / Twilio for voice, Zapier or direct API integrations for plumbing. If you ever wanted to hand it to another engineer, they could read the docs and take it over in an afternoon.
What if my business is too small or too niche?
If you're doing $300k+ in annual revenue and have any kind of sales funnel (even just inbound phone or emails), AI almost certainly fits.
I've built agents for plumbers, ecommerce stores, small law practices, salons, SaaS companies, local services, solo consultants. If you're genuinely too small or too niche — say, a pre-revenue solo operator — I'll tell you straight on the first email. I won't take on a build that isn't the right fit.
How is this different from hiring an AI agency?
Three meaningful differences.
Solo, not layered. No project managers, no discovery calls, no 40-slide deck. You deal with me directly from the first email to the live agent.
Free first build. Agencies structurally can't afford this — their overhead demands they bill upfront. I can, because I'm one person.
Specific, not strategic. I pick one leak to fix, really well. Agencies sell you an "AI transformation" — which works at enterprise scale but is wrong for $500k–$10M businesses. If you're enterprise, hire an agency. If you're me-sized, hire me.
How long before I see ROI?
Usually within the first week of the agent being live — so roughly two and a half weeks from first email.
If the leak was real (missed calls, slow quotes, cold leads), plugging it shows up in the numbers almost immediately. For most clients, the retainer pays for itself in the first week of any given month. The rest is upside.
Will my customers know they're talking to an AI?
Most won't notice. Modern voice AI in 2026 is indistinguishable from a human for the first minute or two of a call.
I don't deceive, though. If a customer asks directly ("am I talking to a real person?"), the agent confirms it's an AI. In practice, customers care far more about whether their issue gets resolved than whether the operator is human. A well-built AI handles both.
What data do you need access to?
The minimum needed to do the job. Usually: calendar access, CRM access, sample past customer interactions (call recordings, emails), and pricing information.
I don't ask for more than that. I sign an NDA if you want one. Your data stays in your systems — the agent reads and writes via standard APIs, and the build documentation shows you exactly what's being accessed and how.
Can I cancel the retainer any time?
Yes. Month-to-month, no lock-in, no exit fees.
If you stop the retainer, the live system continues running — it's yours — but I'm no longer maintaining it. You can keep it running as-is, hand it to someone else, or turn it off. Most clients stay because the ongoing maintenance is genuinely valuable, but the exit door is always open.
What happens if you get hit by a bus?
Everything I build is documented end-to-end. Any capable engineer could pick it up.
There's also a small network of other independent AI operators I know and trust. If something happened to me, I'd refer you to one of them. The system itself doesn't stop working — the tools I use are all standard and keep running regardless of who's maintaining them.
What's the single biggest reason a build doesn't work?
Wrong scope. We picked the wrong leak to fix, or the scope was too ambitious for 14 days.
I try to catch this in the Day 1 scoping call, and I'll push back hard if I think the scope is wrong. If we discover mid-build that we were solving the wrong problem, we pivot. That's happened a handful of times and it's always been fine.
How do I get started?
Email me one sentence about where your business is losing money. That's it. I'll reply within 24 hours with either "yes, let's build" or an honest "no, try this instead."
If you made it this far, you probably already have a specific leak in mind. Don't overthink the email. One sentence. I'll take it from there.