Twelve things people ask before they say yes.

No — ChatGPT is a chatbot you ask, this is infrastructure that runs your business while you sleep. Different category of tool. ChatGPT can't watch your Stripe webhook, route the invoice, draft the customer email, log it to Notion and ping you on Slack only when something looks off. The systems I build do all of that without prompting. They take action, ask when uncertain, and finish the work.

Every automation runs on deterministic scripts — same input, same output. AI handles only the reasoning steps ("is this a real customer?"), plain code handles execution ("send the email, log the lead"). No hallucinations on the doing part. Optional retainer means if anything does break, I'm on it within 24 hours, Brisbane time.

Audits are A$2,000–5,000 (1–2 weeks) and include a written list of what's possible and what each automation would cost to build. Builds are quoted per scope after the audit — no fixed price tag because no two teams have the same workflows. You see the numbers before any build commitment.

Most AI marketing is hype — I agree. That's why I don't sell AI magic. I sell specific automations that save specific hours, with the assumptions written down where you can argue with them. If a workflow shouldn't be automated, I'll tell you and we won't build it. The discovery week exists exactly to separate hype from value before you commit.

Most businesses have more data than they realise — email threads, spreadsheets, the CRM, things people remember. The audit week figures out what's there and what's missing, then writes a build plan around what you actually have. The gap is usually smaller than you think, and the audit gives you that answer before any build commitment.

Fair. The audit (A$2,000–5,000, 1–2 weeks) is built for that — you get a written list of what's possible and what each piece costs, then decide on your own timeline. No pressure to commit to a build at the end. Several clients have sat on the audit for 3–6 months before greenlighting the first build.

Zapier and Make are great for simple if-this-then-that workflows. Axel AI is for the workflows that have decision-making in them — "is this email a real customer enquiry or a vendor pitch?", "does this invoice match a known supplier?", "is this report ready to send or does it need a human eye?". Those branches need reasoning, and reasoning needs an LLM in the loop. The engineering work is in deciding when AI is required and when plain code wins, then wiring the whole thing into your existing tools without the no-code platform sitting in the middle.

A full-time senior AI/automation engineer in Australia runs A$160–220k loaded, plus 3–6 months of recruiting, then ramp time. Axel AI is a fraction of that, scoped to a specific deliverable, with no recruiting risk. The trade-off: you don't get a full-time hire's institutional memory or 40 hours a week of focus. So for ongoing infrastructure beyond the first few automations, the optional retainer (monthly, cancel anytime) covers the gap until in-housing makes sense.

Best fit is teams of 10–100 staff — large enough that workflows are repeating and eating real hours, small enough that there's no automation team in-house. Smaller teams (under 10) are fine if there's one specific workflow chewing through their week. Larger orgs (200+) usually need a vendor with more capacity and procurement to match — Axel AI takes a small handful of clients at a time on purpose.

A written, ranked list of every workflow worth automating, with rough build cost and time-saved estimate per item. Plus a no-build recommendation if any workflow shouldn't be automated. You see the assumptions, you can argue with them, and you decide whether to commission the first build. The audit deliverable is yours to keep regardless of whether you continue — several teams have used it as the brief for a different vendor.

No — and that's the point. Code is committed to your repository (GitHub, GitLab, Bitbucket), infrastructure runs on your accounts (Vercel, AWS, GCP, Modal, Temporal), secrets live in your secret manager. I don't host anything on your behalf from accounts you can't revoke. If the engagement ends, everything keeps running with no further input from me — there's no hidden licence, no recurring tooling fee, no platform that breaks when I'm gone.

Yes — happy to. Anything you share is treated as confidential by default whether or not an NDA is in place, but if your organisation requires one I'll sign yours. For larger orgs an MSA + per-engagement SoW is fine; for most small teams a written email scope is enough. Australian Consumer Law and Queensland jurisdiction apply to engagements unless your MSA specifies otherwise.