Answer

What does the price of an AI project actually depend on?

Four factors, in order of weight: the number of systems that must be connected, whether those systems have a usable interface, how clean your data and documents are, and how many exceptions the process has. The AI model itself is the smallest line. Scope one process and the figure becomes concrete on the first call.

Last updated: 22 September 2026

Connections cost more than intelligence

Reading an email and deciding what it is about is nearly free today. Writing the result into an older ERP that only has a screen, no interface, is the expensive part. Where a system offers a proper interface, a connection is days; where it does not, it is weeks of careful work plus ongoing maintenance.

Exceptions are the hidden multiplier

A process described as one flow usually has five or six special cases that live in someone's head: this customer is invoiced differently, that supplier sends a photo instead of a PDF. Each exception is extra logic and extra testing, and exceptions are the main reason estimates drift.

The practical answer is to automate the 80% that is uniform and route the rest to a person with everything already prepared. That keeps the build small and the result reliable.

What does not move the price much

Volume. Handling 200 emails a day instead of 40 changes the running cost slightly but not the build. This is why automation gets cheaper per item as you grow, and why the same automation stays useful after you double in size.

Asked next

Our systems are old. Does that rule us out?

No. Older systems are usually reachable, either through an interface, a database export or by driving the screens the way a person would. It costs more, so we say so during scoping.

Do we pay for the AI usage separately?

Model and telephony usage are consumption-based and shown as a separate, predictable monthly line, with a cap you set, so a busy month never surprises you.