It is the same AI strategy workshop and AI discovery workshop method we run before our paid build engagements – the difference is that you can stop after the workshop or carry on with us into delivery.
A leading member network, came to us with a knowledge chatbot problem. Off-the-shelf OpenAI Assistants with file search had been unreliable and accuracy was critical. In a focused use case workshop we mapped their data sources, scored the candidate against impact, feasibility, data and risk, and produced a four-sprint scope: n8n in Berlin for orchestration, Qdrant in the EU for vector search, GPT-4 via Microsoft EU sovereignty as the LLM, with HumHub integration. Version 1 is now in production serving more than 1,000 network members, time-aware, and operated by a non-developer team. The same scoring pattern shaped our work for the German Insurers Association (GDV) across tens of thousands of policy documents for 400+ member companies, and a Civic Coding-funded AI transcription pilot for Tannenhof Berlin-Brandenburg on EU-hosted infrastructure.
Real process mapping
Defensible scoring
Buildable project sketch
A ranked use case shortlist
defensible against four criteria: impact, feasibility, data, risk
A project sketch
for the top use case, ready to move into a four-sprint build
An effort estimate
you can take into a budget conversation with your CFO
Team alignment
a shared vocabulary across decision-makers and the team that does the work
A 48-hour written summary
shortlist, project sketch and recommended next step in writing
1. Map and surface
2. Score and prioritise
3. Sketch and plan
We bring a decade of impact-tech experience and more than 160 AI projects since 2019. We do not run inspiration days. We run scoping sessions and build engagements that ship, the way we have delivered AI for the organisations below:
A workshop is not a substitute for a real prototype. If you need proof that AI can do the job before committing budget, the cleanest path is to leave the workshop with a four-sprint scope and start building – the way a leading member network did.
AI does not fit everywhere. Rule-based automation already works for some processes. For others, the cost of a wrong AI answer cannot be tolerated even with human review. For both, the honest answer is "not now, and here is what to do instead".
If your data is not ready, no workshop will fix that. The answer is usually a smaller, less glamorous data project before any AI – not more workshops.
How is this different from a generic AI strategy workshop?
How long does the AI discovery workshop take?
What does it cost?
Do we need data to be ready?
What happens after the workshop?
Can you build the prototype too?
Tell us where you are stuck. We will reply with a proposal and a date, usually within a working day.
Simon Stegemann
Co-Founder and CEO
