Five stages, each with a defined output and a sign-off. Nothing moves to the next stage until you have approved the one before it, and the first stage is fixed price so you can stop after it without having lost much.
One workflow, one week, fixed price. A written answer on whether automation pays, and what it costs to run.
One workflow is profiled end to end: the happy path, the exceptions, the data behind it and what it costs to run today. You get a written recommendation - build, buy or stop - with the reasoning attached, so it can be challenged rather than taken on trust.
Time from you: two sessions with the people who actually run the workflow, not only the people who own it.
The design: data, services, interfaces and the failure modes your team will have to operate around.
The design is written against the team that will maintain it, not against the most interesting architecture available. Services, data, interfaces and the failure modes each one introduces, with the trade-offs recorded rather than settled quietly.
Time from you: a technical review with whoever will own this after I have gone.
Hands-on implementation alongside your engineers, rather than a specification handed over a wall.
Implementation happens alongside your engineers, in your repositories, against your review process. A specification handed over a wall is how a system arrives that nobody in the building understands.
Time from you: engineering time in the same sprints, and access that does not take six weeks to arrange.
Runbooks, documentation and every failure traced to its layer, so the system outlives the engagement.
Runbooks, documentation, and every failure traced to the layer it belongs to rather than patched where it surfaced. The test of a handover is whether your team can change the system next quarter without me.
Time from you: a named owner, and one person who has to be able to run it unaided before sign-off.
A scheduled look back at what the system actually cost and what it actually returned.
A scheduled look back at what the system actually cost to run against what was estimated, and what it actually returned against what was claimed. Including the cases where the honest answer is that it did not pay.
Time from you: access to the running costs, and a willingness to hear the result.
The four things enterprise and public sector buyers ask before anything is signed.
I hold no reseller agreements with any vendor or cloud provider, and take no commission on anything I recommend. The recommendation is the same whether you build it, buy it, or decide to do neither - which is the only position from which "do not build this" is worth hearing.
Full assignment of copyright on delivery, with source, infrastructure definitions and documentation handed over. No licence fees, no hosted component only I can operate, and no dependency on me to make a change.
Fixed scope, a written statement of work, stage sign-offs and the usual supplier due-diligence pack. I can respond to an ITT or a framework mini-competition in the format your team needs.
The honest risk with a single consultant is that there is one of me. So nothing is built that only I can operate, the handover is a deliverable rather than a courtesy, and every engagement names in writing who picks the work up if I cannot.
One workflow. One week. A fixed price, and a written answer you can put in front of a board - including the answer that says do not build it.