After the first scoping meeting, we tell you which target we will measure the pilot against, and what it costs.
First step · AI Pilot
We test AI on your own data, against a target set in advance
How many of your pilots went live?
If there is no ready-made solution on the market for the problem, we build a trial system for a single task on the company’s own data, and measure whether it reaches the target set. This shows whether it works before the company spends a larger sum on it. And you take a measured result to management, not another stalled pilot.
Who it is for, and when it is worth it
If the problem is unique, and you do not want to start development blind
For example, a special document type, a production phenomenon or a forecasting task for which there is no ready-made product.
Who it is for
Heads of operations, CIOs and plant directors, typically after a Situation Assessment or an AI Compass Audit.
When it is worth it
When a quarterly or annual budget for experiments opens up that has to be used, and management expects measured results.
What we need from you
A completed assessment, access to the company’s own data and a measurable goal.
What it is not for
We do not recommend it for a problem that already has a ready-made market solution. Nor is it built for production use: the pilot prepares the decision.
Does this sound familiar?
When the pilot succeeds, and then nothing comes of it
This is what we hear from digital and operations leaders.
The failure
The pilot stalls before going live
“Another stalled pilot counts against me.”
Head of innovation and digital, large enterprise
Measurability
There is no measurable result
“Not another programme that brings no results.”
Head of operations, large enterprise
In front of management
There is no time for a long development
“Management expects visible results.”
Head of innovation and digital, large enterprise
In front of the group
Nobody measures the vendors’ promises
“I have to show the group measured results.”
Head of operations, large enterprise
Cost
You do not know what the production system will cost
“I want to know in advance what it will cost if we take it live.”
Head of innovation and digital, large enterprise
Time
The data is late, and the budget expires
“It has to fit into one quarter.”
Head of innovation and digital, large enterprise
Does this sound familiar?
If you recognise yourself, let’s look together at what is worth testing on your own data.
What you get at the end
A measured result and a decision, not just a demo
We set the target at the start, and at the end we measure the result against it.
- 01A working trial system on your own data
- 02A measured result against the target set
- 03A decision report: is it worth developing into a production system, and what would it cost
How it runs
4-12 weeks, with a decision point set in advance
01
Target
We agree in advance on what result counts as working.
02
Data
We get access to the data on which the trial system runs.
03
Build
Depending on the problem, AI development, document processing, image and video analysis, or a forecasting model.
04
Measurement
We measure the result against the target set.
05
Decision
It becomes clear whether it is worth taking live, and what it would cost. The decision is yours.
Price and success criterion
A price after the first scoping meeting, with a success criterion set in advance
We give a price once we know the problem and the data. The pilot is not free of charge, because we are responsible for the result. If it becomes a production system, we deduct the pilot fee from the price of the production system.
Frequently asked questions
What leaders usually ask about the AI Pilot
How is it different from a vendor’s free trial project?
We work with a measurable goal and a decision point set in advance. It is not free of charge, because we are responsible for the result.
What is needed to get started?
You need a completed assessment (Situation Assessment or AI Compass Audit), access to the company’s data and a measurable goal.
How long does it take?
4-12 weeks, depending on the problem and the data.
What happens if the pilot does not reach the target?
The decision report records this too, so the company does not spend a larger sum on a development that does not work on its own data.
Do you have references?
We publish named cases with the client’s permission.
Before you decide
The AI Pilot product sheet
What we did not cover on this page: the steps from a small first step to the complete solution, why it pays to go this way, and how we work along the way. Feel free to forward it to your colleagues.
Not the right fit? You can also start this way
AI Compass Audit
If you first need to decide where AI is worth introducing, and in what order.
Situation Assessment
If you would first like to assess where most time goes on manual work. HUF 199,000 + VAT.
Solutions
If there is a ready-made solution to the problem, you do not need a pilot.
FURTHER READING
Why do most pilots stall?
These articles show what turns a trial into a production system, and what does not.
Just looking around?
Read how a pilot becomes a production system, or subscribe to the Signal over Hype newsletter.
First meeting
Request a first meeting
Briefly describe the problem and what data you have for it. We give you a price after the meeting.
Prefer the phone? +36 30 235 0100