Five fixed-scope services. Written down before you pay.

You do, on full payment. The workflows, the prompts, the evaluation suite, the runbook and the source all live in your repositories and your infrastructure from day one. I hold delegated admin access during the engagement, which you can revoke at any time. There is no vendor lock-in to unwind, because there is no vendor tenancy to be locked into.

Everything runs in your own tenancy — your cloud account, your automation instance, your vector database, your model API keys. That is deliberate on three grounds: you never lose access to your own operations if we stop working together, you see the real running cost directly rather than through my markup, and it keeps the arrangement clean under the licences of the tools involved.

Model and third-party SaaS costs are billed to your own accounts and are never resold through me — I do not want to carry token-price risk on your behalf and you should not pay a margin on it. Every AI build ships with a cost-per-conversation or cost-per-run projection so the monthly number is known before launch rather than discovered afterwards. Ongoing support is optional, quoted monthly against the tier you choose, and never bundled into the build price.

Every build ships with retries, timeouts, dead-letter handling and failure alerting into a channel your team already reads, plus a written runbook. There is a 14-day bug window after handover at no cost. Beyond that, a Care retainer covers monitoring and fixes with a response SLA of two business days, one business day, or four business hours depending on tier. Without a retainer you can still call me — it is quoted as ad-hoc engineering.

Every deliverable that touches a language model ships with an evaluation suite: at minimum forty golden question-and-answer pairs, a documented pass threshold, and a regression run before handover and before any subsequent change. It means "the assistant got worse" becomes a number you can check rather than an argument. Very little at this price point ships with one, and it is the main reason to choose this over a cheaper prototype.

Only what the workflow you approve has to send, to the model providers you have chosen, from infrastructure you control. I will give you a written one-page summary before any build starts covering which sub-processors are involved, what is sent to each, where it is stored, how long it is retained, and how it is deleted. Where a provider offers a no-training-on-your-data setting, that is the configuration used by default.

It is quoted per engagement rather than published, because the same brief can differ by a factor of three depending on your systems. Six things decide it: how many systems it touches, whether those systems have a usable API, how clean the data is, how much of the process is already written down, the corpus size and channel for an agent, and whether the output is customer-facing or internal. You get a fixed written price after a twenty-minute call, usually within twenty-four hours of it, and that price does not move afterwards — two revision rounds are included and anything outside the written scope caps is agreed before it happens.

Because a published number would have to be padded to cover the worst case, which means straightforward projects subsidise difficult ones. The alternative — advertising a low number and discovering the gap in week two — is how fixed-price projects turn into arguments. Instead the entire commercial structure is published: what drives the number, what is included at no extra charge, the payment milestones, the change-order rule, and the tax treatment. Only the amount waits for a short conversation about your actual situation.

Every enquiry gets a reply within one business day, from the person who would do the work rather than an automated sequence. After a twenty-minute fit call, a written fixed-price proposal typically follows within twenty-four hours, containing the scope caps, deliverables, timeline and payment schedule. There is no multi-week procurement dance, because there is no sales team to route it through.

That is the expected end state, and the build is structured for it. Everything is in version control with a runbook, prompts are versioned files rather than text pasted into a web interface, and the handover session is recorded. Your engineer can pick it up by reading the repository. There is no proprietary layer of mine in the middle that has to be replaced first.

Numeric caps written into the statement of work before you pay — how many processes, how many integrations, how many revision rounds, how many documents. Two revision rounds are included. Anything beyond the caps is a change order, quoted and agreed in writing before any work happens, at the rate stated in your statement of work. The fixed price does not move because the scope does not move quietly.

It is a fair question and the honest answer shapes how everything is built. Because the work lives in your infrastructure, in your version control, with a runbook and an evaluation suite, any competent engineer can take over without me — that is the point of building it this way rather than faster and looser. Delivery capacity is deliberately capped so schedules are not built on optimism, and for larger engagements I will name trusted subcontractors in advance rather than after a problem.