Skip to content

AI Pilot

Test an AI feature on your own data before you build all of it

AI Pilot is a service for companies that want to add AI to their product or process (a support agent, document search, processing of incoming requests or an AI feature in an existing app) but do not yet know whether it will work well enough. In 3 weeks we build a working pilot on your real data, measure how often it answers correctly and what one use costs, and tell you whether it is worth taking to production.

  • A working pilot on your data - something your people try on their own cases, not a presentation.

  • Measured quality and cost - you know how often the AI answers correctly and what one use costs.

  • A decision with a price - go to production, change it or stop, with an estimate of cost and timeline.

A first version is quick to build. The question is whether it works on your data.

A first version of an AI feature can be built in a few days today. It is harder to know whether it will hold up once real users get it: whether it answers from your sources, what it does when it does not know the answer, and what it costs at a thousand uses a day.

So we build the pilot narrow, around one use case, and measure it on questions and documents you prepare. If the numbers show that AI does not pay off in this place, we say so and save you the full build.

What we can test

  • An AI support agent. It answers employees or customers from your knowledge base and says when it has no answer.

  • Document search. Contracts, manuals, technical documentation: an answer with its source.

  • An AI feature in an existing product. Suggestions, sorting or auto-fill where it really saves the user time.

  • Automation of repetitive work. Processing incoming requests, email or forms, with human review where it is needed.

How it works

We start with a free intro call, where we pick one use case and agree how we will measure success. We sign a non-disclosure agreement and you prepare the data and the test cases. A developer and a UX designer build the pilot in a test environment, your team tries it, and at the end you get a report with the measurements and a presentation.

If you decide to go to production, we extend the pilot in Build & Launch or develop it step by step with the On-Demand Product Team. If you already have an AI solution and it does not work as it should, the right start is AI Project Rescue.

✅ Ideal for companies that have a concrete use case and the data for it, and want proof that it works before approving the budget for the full build.

When?

When you have a concrete use case for AI and the data for it, and management wants to see it working before approving the budget.

Why?

Because with AI, quality only shows on your own data. A pilot costs a fraction of the full build and tells you whether it is worth starting, what it will cost to run and where the risks are.

Answer quality: how often the answer is correct on your test cases and what the system does when it does not know.

Cost: the model cost per use and an estimate of the monthly cost at the expected volume.

Data and privacy: which data goes to the AI model provider, where it is stored, GDPR compliance.

Security: who can access what and how the system responds to attempts at misuse.

User experience: whether the user knows when to trust an answer and when a person has to decide.

Path to production: what still has to be done for the pilot to become part of your product or process.

Ready for a first pilot?

Describe what you would like to test. We reply within one working day.

From 5.500 €

Duration: 3 weeks
Taxes (VAT or similar) are not included.

What would you like to test with AI?

In a few sentences, describe the use case and the data you have. We will look together at whether a pilot is the right step. If it is not, we will tell you.

All fields are required

Explore other service packages

Is the idea worth building?
Concept Review

In 2–4 weeks you know whether the idea is worth building, what to change about it and roughly what it would cost.

What exactly do we build, and for how much?
Product Discovery

In 2–4 weeks you get a prototype that users have tested, a prioritised feature list and a cost estimate for development.

EAA compliance
Accessibility Audit

An audit against WCAG 2.1/2.2 AA for compliance with the European Accessibility Act. In 2–3 weeks you get a list of barriers, annotated screenshots and instructions for developers.

A team for a live product
On-Demand Product Team

A monthly subscription to UX designers, developers and a project manager. One task at a time, an unlimited queue, work starts within 48 hours.

From 4.500 € / month

Explore service details
From plan to launch
Build & Launch

Design, development, testing and launch of a digital product with a senior team. The code and the design are yours at launch.

Estimated per project

Explore service details
When an AI project stalls
AI Project Rescue

A prototype from an AI tool works in the demo but not in real life. In 2 weeks you learn what is wrong in the code, what is worth keeping, and what it costs to get the product to users.