Service A · AI consulting & delivery
The AI studio that measures.
We find the tasks where AI truly helps, build a prototype on your data in five days – and tell you honestly how good it is.
01Why companies bring us in
Less manual work, more judgement.
Screen mountains of data
Images, maps, documents: AI pre-sorts, people decide on a short, reasoned list.
14,600 tiles in 12 min
Digitise processes
Paper forms, inspection records, quotes – as an app that captures data cleanly and generates documents itself.
PPE inspection app in 5 days
Make knowledge accessible
Assistants that answer from your documents and say when they don’t know.
Live on this site: ⌘K
Verify AI results
Already using AI? We independently measure how good it really is – blind and with strict counting.
95.6 % → 88.8 % recounted
02The AI sprint
Five days. Then you know.
Instead of months of concept work, a working prototype on your real data. On Friday you get numbers: how often the AI is right, where it errs, what operation costs.
Mon · Day 1
Goal & data
Sharpen the question, review data, define success
Tue · Day 2
Pipeline
First end-to-end run on real data
Wed · Day 3
Results
Study error patterns, refine
Thu · Day 4
Measurement
Blind test against a sample
Fri · Day 5
Decision
Handover, numbers, clear recommendation
03How we verify
Trust is measurable.
The reviewing instance knows neither location nor expectation. Every rejection is reasoned and cross-checked.
How much that matters shows in a re-evaluation: for a published open-source workflow detecting archaeological structures, our strict count gave 88.8 % instead of the reported 95.6 %. No accusation – counting methods make the difference, and we make it visible.
- #0417Straight edge, 180 mrejected · forest road, aerial 1968 ✓ cross-checked
- #0418Round mound, 14 mopen · survey recommended
- #0421Parallel furrowsrejected · skid trails ✓ cross-checked
- #0426Rectangular hollowrejected · timber yard, map 1987 ✓ cross-checked
- #0430Edge on slopeexplained · vineyard terrace, 19th c. ✓ cross-checked
04Case studies
Fast and evidenced.
< 2 days · > 1,000 km²
Satellite data pilot
From brief to prioritised list in under two days – AI screening, blind review, chain of evidence.
Read the case study
5 days
Digital PPE inspection record
A paper form for fall-protection equipment became an app with QR codes, signatures and PDFs.
Read the case study
06Formats
Start small, scale cleanly.
30 minutes, free
Intro call
Your problem, our honest view on whether AI solves it.
One day · fixed price
AI opportunity analysis
Prioritised use cases with effort, benefit and risks as a written roadmap.
Five days · fixed price
AI sprint
A working prototype on your data, measured against a sample.
Fixed price
AI audit
An existing AI workflow, blind re-measured, with a report on its error profile.
Quote after the sprint
Delivery
Workflows, agents and integration with your systems – production-ready and tested.
Monthly
Care
Operation, monitoring, adjustments. Model costs transparent by usage.
07Fragen
Frequently asked.
What does AI consulting with headon cost?
Entry formats and the sprint have fixed prices; delivery is quoted after the sprint – when both sides know what they are talking about. All prices are on the pricing page.
Do we need our own data?
For a meaningful sprint, yes: a realistic sample is enough. We help put it together and say early if the data will not carry.
What happens to our data?
It stays yours. We process only what the project needs and can run confidential data through local models on our own hardware.
What if AI doesn’t solve our problem?
Then we say so – after the sprint at the latest, with numbers. A reasoned “no” saves you more than a half-baked project.
Do you work internationally?
Yes. We are based in Germany and work remotely with companies, public bodies and research teams across Europe and beyond.
Which task should AI take over?
Describe the work that eats time today. In the intro call we tell you honestly whether AI helps – and how to find out in five days.