The things people actually ask before they say yes.

Nothing, and thirty minutes. There is no deck.

Most people asking this have been pitched before and are checking whether this is another one. It is not — we ask what eats your week and tell you whether there is anything in it worth automating at all.

If there is not, that is a perfectly good outcome and you will hear it on that call rather than after you have paid for an audit to find out.

Fair question, and the honest version of it is “what stops this being five hundred dollars for a PDF?”

Ninety minutes on your real week — not a questionnaire, we follow the work through whatever tools it already lives in — and then a written report.

The report lists every automation candidate we found, ranked by payback: how many hours a month it gives back, a fixed build quote, and a feasibility check. The ones that are not worth building are in there too, with the reason.

You own the report either way. Build one candidate, build three, or take the whole thing to another firm and have them quote against it. The $500 is credited if you build with us.

Then we say so and that is the end of it. You keep the report, which is now a written map of your own process — something most businesses have never had.

An audit that always finds work is a sales call. We would rather lose the build than sell you one that does not pay for itself.

• The call — thirty minutes.
• The audit — ninety minutes with you, then the written report.
• The build — most go live two to four weeks after you approve the quote.
• Shadow testing — five to seven days of that, running against your live data before it touches anything.

Some take months, and we would rather say so here than in week six. Usually that is scope — one workflow is not three — or a system that will not give up its data cleanly. A workflow reading a tidy API is not the same job as one reading a platform with no usable export.

The feasibility check in the audit is where you find out which of those you have, and it happens before you have committed to anything.

The work that is repetitive, has a rule behind it, and currently depends on a person remembering: reading and routing incoming email, answering the questions that have a factual answer, re-keying data between systems with verification against the source, chasing quotes and documents on a schedule, and assembling paperwork consistently.

The categories are listed in full, but most of what we build does not fit one neatly. Odd platform, odd process, a workflow that grew around how your business actually works. That is normal, and it is what the audit is for.

Three kinds of thing, and we will tell you about all three during the audit.

• Anything that has to hold up in court or in a relationship. Judgement, negotiation, bad news, and the message that decides whether someone stays a customer.
• Anything touching money. Refunds and payments are prepared and left for a person. Our builds have no ability to send or pay at all.
• Anything sitting on data we cannot reach properly. If your system has no usable export, automating it means scraping a screen, and it will break the first time the screen changes. We put that in the report as not viable rather than quoting for it.

No. The whole approach is the opposite of that: we build around whatever you already run, in accounts you already own.

Replacing your systems would mean retraining your team, migrating your data, and betting your operations on our choice of tools. None of that is automation. It is a software project with automation attached.

A person approves everything that matters, and not because we configured it that way. The builds do not contain the ability to send a message or move money. The node is absent from the workflow.

So the system reads, files, labels, drafts and escalates. Anything that would reach a customer or touch a payment is prepared, evidenced, and left for you to action in your own tools. A filter can be bypassed; a missing tool cannot. The full picture is on the security page.

The narrowest access that lets the work happen, scoped per workflow rather than per account. A build that reads your inbox does not get your billing.

We never ask for a password. Access is added the way your platform adds a collaborator, from your side, visible in your own admin log — so you can see exactly what we hold at any time without asking us.

You do, and you always did. The workflows, the prompts, the logic and the documentation live in your accounts and are yours to keep, edit, or hand to someone else.

Nothing switches off when you cancel. There is no licence key to expire and no service of ours to disconnect — the system keeps running with nobody on retainer.

Typically $10–50 a month, billed directly in your own account. You pay the provider what the provider charges.

We never mark it up and never take a margin on it. That is the machine, not us — the whole fee schedule is published, including what we do charge for and why.

First, we find out. On our own systems 4 of 27 scheduled runs once left no record. We checked two and found real errors nobody was alerted to; the other two, we still don’t know. Every monitor we had watched for failures, not for absence.

So every build we ship alerts on the run that never happens, not only the run that errors. That record is published, including the number that does not flatter us.

After that it is an escalation path with published response targets by severity, not a favour. They are on the pricing page.

Three things priced separately, not a bucket of hours: keeping it running, a fixed number of build days each month to improve it, and someone on call when it breaks.

Plus one page every month in the same format: what was patched, what shipped, what broke, what it cost, and one recommendation for next month.

If your build days go unused three months running, we will propose downsizing you to monitoring and on-call only. Both tiers are published.

Cancel with thirty days for any reason. No annual lock-in and no exit interview — if the system no longer needs us, that is the automation working the way it was supposed to.

Or pause for up to sixty days a year at a quarter of the retainer, with monitoring still on, so a quiet quarter does not force you to cancel and rebuild later.

It archives and labels rather than deleting, so anything it touched in the last thirty days can be put back. A mistake should cost minutes, not data.

That describes how the automation behaves inside your own platform. It is not a backup service, and we would not want you treating it as one.