The aftercare question
Consulting, build, or build-and-run — the wrong depth wastes more money than the wrong model.
Buyers often ask for “an AI project.” What they need is a depth decision: clarity only, clarity plus a system, or a system someone will keep alive.
We ask this early because it changes the work. A roadmap is not a deployment. A deployment is not a Tuesday when the integration stops writing back to the CRM.
When consulting is enough
Consulting alone is right when the organisation still does not know where AI pays — or where it should not. The output is a clearer yes, no, or not yet. It is also right when the process is too undefined to automate, and building now would only freeze the confusion.
If you want a first-call shape for that, read bring one process.
When a build is enough
Build is right when the process is clear, ownership exists, and your team can absorb the system. That includes the boring parts: who watches exceptions, who updates the approved answers, who gets the alert when a write-back fails.
Build-and-transfer is a real finish. We design, prove, hand over with controls and training, and step back. It only works if the desk on the other side of the handoff is actually staffed for that.
When someone has to stay
Aftercare is right when the desk cannot absorb another orphaned integration. Servers, prompts, monitoring, and exception review are part of the product. If nobody will own those, a pilot is a liability with a launch date.
This is not a luxury upsell. It is the admission that software keeps needing a person. Managed operations is that line of work: we run under agreed outcomes, including the unglamorous part where someone still answers when the server misbehaves.
How we work, in four verbs
Diagnose, engineer, prove, deliver. Aftercare sits in deliver — and only when you asked for it. See how we work for the sequence, then decide the depth before you decide the stack.
Keep reading.
Bring one process to the first conversation
A useful first call is not a tour of models. It is one process, the depth you think you need, and an honest list of what you do not know yet.
Read the note
Start with the work. Not the AI.
The useful first question is not which model to buy. It is which process still depends on memory, retyping, and whoever happens to be at the desk.
Read the note
Demos aren't desks
An agent that dazzles in a slide deck still has to survive your exceptions, approvals, and Tuesday afternoon volume.
Read the note