Text Link
1/5

Desired Service

PROJECT DETAILS

2/5

BUDGET

3/5

TIMEFRAME

4/5

CONTACT DETAILS

We use your details solely to process your enquiry. Details in our privacy policy.

5/5

THANK YOU

Our AI agent was quick: the reply to your initial enquiry is already in your inbox.

ERROR

THANK YOU!

We will get back to you shortly.

AI pilot · From PoC to production

The proof of concept worked. Then nothing moved.

The report is written, the board has seen it. A year later none of it is in use. The technology is rarely the problem. The reasons an AI pilot stalls are few, and they are almost always baked into the first design.

Six reasons, one patternSeven stepsOther vendors’ pilots tooFirst call free of charge
94 %
Less effort
23 h down to 84 min per event
7
Steps
an exit at the end of each
2 weeks
To the concept
with a fixed price
from 5,000 EUR
Project size
fixed price, in writing
PoC (proof of concept)
A proof of concept, or PoC, checks on a small scale whether an idea can be built. Production asks for different things: real data with its exceptions, a named owner, integration with the systems already in place. AI pilots stall in that gap.
01 Reasons

Why pilots stall

None of these show up during the pilot. It is set up so they do not.

01

Clean data in the test

For the pilot someone picks the ten best documents. In production you get scans from 2014, three versions of the same template and tables with empty columns. Accuracy drops, and the discussion starts.

Data realityExceptions
02

Nobody runs it

Who updates the index, who takes the questions, who notices when quality slips? Between IT and the business unit that stays open. The system is not switched off. It just stops being used.

OwnershipHandover
03

No approvals built in

What happens on an obviously wrong output? Who stops it, where is it confirmed? Until that is settled nobody signs the request for production. Many pilots hang on exactly this point, not on the technology.

Stop ruleApproval
04

Cost per request unknown

The pilot ran 40 requests a day on a developer’s account. What 4,000 requests cost, which model still carries that and who gets the invoice, nobody has worked out.

Model costBudget
05

Users were not in the room

The pilot was built for the business unit, not with it. The people who are supposed to use it every day see it for the first time in the closing presentation. They find the three cases that fail within five minutes.

Business unitReview
06

Security sign-off missing

The pilot ran on a laptop with a test file. Production needs data protection, IT security and often the works council to agree. Start that after the pilot and you lose months.

Data protectionIT security

All six can be settled before the pilot. Afterwards it is tedious. Beforehand it costs almost nothing.

88 % use it, a third scale it
88 % of the organisations surveyed use AI regularly in at least one business function. Almost two thirds have not yet begun rolling it out across the company, and only 39 % report any effect on operating profit at all. The gap between trying and running is exactly what this page is about.
Source: McKinsey, The state of AI in 2025, Global Survey, 5 November 2025, 1,993 respondents in 105 countries. mckinsey.com
02 Triage

Continue, re-test or shut down

Not every stalled pilot needs to run again. We sort by day-to-day impact and by how certain the implementation is.

Impact high · certainty high

Straight to production

Only integration and an owner are missing. That takes a few weeks. This is where we start.

Impact high · certainty low

Re-test first

Run it on messy data and see where quality breaks. Design comes after that, not before.

Impact low · certainty high

Later

It clearly can be built. There is no reason to hurry, so it moves down the list.

Impact low · certainty low

Shut down

Whatever lands here gets closed. That is part of the job too.

03 Comparison

Pilot versus production

Same idea, two different sets of demands. The right-hand column is what a proof of concept usually does not deliver.

CriterionIn the pilotIn production
DataTen hand-picked documentsEverything on the file server, exceptions included
Success metric“It works”Hours that measurably disappear
IntegrationData pasted in by handInterfaces to ERP, document store, mailbox
OwnerThe developer who built the pilotA named person in the business unit
Error caseClicked awayStop rule and approval, documented and built in
CostTest accountCost per request known, budget approved
Sign-offNoneData protection, IT security, works council if needed
04 Process

Seven steps, an exit at the end of each

At the end of every step you decide whether to go on. If you stop, you keep everything produced up to that point.

The steps are separate so you can get out. Order everything at once and people stop raising the awkward findings halfway through.

Step 1

Enquiry

Where is it stuck? We read the old pilot’s material, even if another vendor built it.

Step 2

Consultation

Triage by impact and certainty. Set the order, name the candidates for shutting down.

Step 3

Concept

Success metric in numbers, stop rule and owner go in writing here, together with the fixed price.

Step 4

Development

Built on real data, exceptions included. Integration with your systems comes first, not last.

Step 5

Review

The business unit uses the system and lists what does not fit. We work through the list.

Step 6

Acceptance

Checked against the metric from step 3. It is not moved afterwards.

Step 7

Operation

Handbook, monitoring, handover to your named person. We stay on under a maintenance contract if you want.

05 Preparation

What belongs in the pilot already

Six decisions that rarely get made in a proof of concept. They are exactly what is missing later when production is requested.

Make them at the start and they cost almost nothing. Catch up on them at the end and you pay in months.

Success metric
Not “looks usable” but how many hours have to disappear for it to continue. The number is not moved afterwards.
Messy data
Not the prepared sample documents but the clutter from the business unit, from day one. That way you know early where quality breaks.
Stop rule
Who stops an obviously wrong output, and at which point? That is not just on paper, it is built into the system.
Owner
Not the department, a person with a name. They join from the middle of development.
Integration trial
The connection to ERP or the document store is tried once during the pilot. That effort is the hardest to estimate.
Handover
Source code, handbook, update path. The goal is a system that runs without us.
06 Evidence

Three published projects, all in production

We do not show pilots as references. Only systems that have been handed over and run at the client appear on this site.

Each has a case study with architecture and numbers.

Meeting dashboard
Processes meetings daily, from transcript to task assignment. Read the case study
Knowledge agent
Searches across Japanese and Thai material. Self test 0.9 of 1.0 across 100 scenarios, in production. Read the case study
Élysée Events
Runs on real events. 23 hours down to 84 minutes per event, 94 % less effort. Read the case study
07 Self check

Five questions about your pilot

If you cannot answer three of them straight away, it is worth going back to the concept.

Question 1
What had to drop, and by how much, for the pilot to count as a success? Was that number fixed beforehand?
Question 2
Did the test run on everyday data or on prepared examples?
Question 3
Is the person who runs the system named? Not the department, the person.
Question 4
Was the system connected to ERP, document store or mailbox, or was data pasted in by hand?
Question 5
Who stops a wrong output, and is that procedure written down anywhere?
08 Frequently asked questions

From pilot to production: frequently asked questions

Another vendor built the pilot. Will you take it over?+
Yes, that is actually the more common case. We read the report and, where possible, the source code. Whether a rebuild is needed depends on what is in there. At the very least we then know what was already tested and do not repeat it.
Does the proof of concept have to be repeated?+
Usually not. That the technology works has been shown. What is missing is almost always the same four things: a success metric in numbers, a test on messy data, integration with existing systems and a person who takes over operation. Adding those is faster than a second pilot.
Where do you start?+
With a list of every stalled initiative, sorted by day-to-day impact and by implementation certainty. Not all of them need to run. Whatever brings little and is uncertain, we propose to shut down.
We have nobody who can run a system like this.+
The person who will look after it joins from the middle of development. There is a handbook at handover. If that is not enough we run it under a maintenance contract, but the goal stays that you can do it yourselves. A system that only runs with us stalls again at the next change.
How long until it is in production?+
If only integration and ownership are missing, a few weeks. If the concept has to be redone, months. After the first call we tell you which case you are in.
What happens if the effect does not show?+
If the metric from the concept is not reached we say so and stop. The metric is not moved afterwards. Otherwise there is always a reason to keep going.
What about our existing IT provider?+
We replace nobody. Whoever looks after ERP or the network keeps doing that. The split is written into the concept. Usually it is enough for us to take the AI part.
09 Read on

Related topics

Next step

Show us the pilot, exactly as it is.

In thirty minutes we can tell you whether it is worth carrying on. If it is not, we say that too.

DEJPEN