What it costs, and why the honest answer starts with a question.

We do not publish a price list because it would be false. We publish the estimating method and the ranges we work in — which is what lets you decide whether a conversation is worth booking.

Why there is no table

Two projects with the same name — “assistant over internal documents” — can differ tenfold in effort. What changes is not the model: it is the state of the data, the number of systems to connect, who has to approve what, and whether someone on your side is available to answer questions while the work happens.

A table that ignores that is either inflated to cover the worst case, or understated and corrected halfway through. We prefer to estimate after we understand.

The ranges we work in

• Opportunity assessment. Two to four weeks. A fixed figure, known before we start, and the cheapest work we do — deliberately, because it is what stops the rest being spent badly.
• First case in production (assistant, automation or agent). Six to twelve weeks of work. Fixed price once the scope is settled; iterative while it is not.
• Integration with existing systems. Almost entirely dependent on how many systems there are and how good their APIs turn out to be. This is the line where estimating without a survey is least reliable.
• Governance and audit. Bounded work, with the documentary deliverable defined up front.
• Monthly running. Model consumption, hosting and maintenance. We instrument per process from day one so this number is predictable rather than a surprise.

What pushes the cost up

• Data scattered across systems that do not talk to each other, or with no reference version.
• Legacy systems without an API, where the connection must be built before any AI work starts.
• Compliance requirements that force a self-hosted model instead of a commercial API.
• No single person on your side with the time and authority to decide.

What brings the cost down

• Starting with one measured case rather than a platform.
• Data already centralised, even if untidy.
• Accepting human review in the first phase instead of demanding full automation from the start.
• Having the assessment done — it reduces the uncertainty in the estimate, and therefore the risk margin it has to carry.

How we contract

Stable scope, fixed price. Scope still being discovered, iterations with priorities reviewed and the option to stop at the end of each one. We do not write contracts that force you to continue in order to justify what has already been spent.

Frequently asked questions

What is the smallest project you take on?

The assessment. It is the natural entry point and commits you to nothing afterwards. Below that, the right answer is usually that it is not yet time to hire anyone.

Will you estimate without a meeting?

We will give an order of magnitude by email if the description is concrete enough. A real estimate requires understanding the process, and that takes two conversations.

Do you charge for the assessment if we go ahead with you?

The assessment is standalone work with its own value and is invoiced as such. The report is yours and can be used to hire anyone, including another firm.

← All services