Something real to click, running in the cloud
Not a mockup and not a slide. Software on your data, in your AWS account, that a sceptical colleague can open and try to break. Usually a matter of hours, not months.
- AI Prototype Workshop
- €7,000
- 2–24 hours of build
- Your AWS account
The first rung where the idea can lose
Everything before this can be argued with. A demo cannot. It either handles the awkward record or it does not, and the room finds out together.
That is the whole point of putting it this early and building it this small. A demo is deliberately not production: no hardening, no scale, no operations story. It answers one question, which is whether the approach survives contact with real data.
It also changes the conversation. Sponsors who could not be pinned down on a document will tell you precisely what is wrong with a demo in ninety seconds.
Working software and an honest measurement
Built in the AI Prototype Workshop: kick-off, the build in between, and a decision session at the end.
A working prototypeIncluded
Running on your real data in your AWS account, with the code in your repository from day one.
Measured evaluationIncluded
Accuracy, latency and cost per request against the success criteria agreed at kick-off, not against a demo dataset.
Go or stop decision recordIncluded
The evidence, the trade-offs and the recommended next step, written for people who were not in the room.
Three things a demo cannot tell you
That it scales
Ten documents working says nothing about ten thousand. That is the next rung's question.
That people will use it
Colleagues trying a demo are not customers doing their job under time pressure.
That it is compliant
Obligations under the EU AI Act depend on the finished system and its risk class[8], not on the prototype.
From demo to first real users
If the demo holds, the next rung puts it in front of people who did not build it and are not being polite. That is the MVP or funded proof of concept, and it is the first rung with a real timeline attached.
Ideas
Who needs this most?
The idea said out loud, and who it is actually for. Out: a shortlist
Free 30-min callClarity
Is the idea clear?
Written so anyone in the room can repeat it back without you. Out: one page
Concept
Is it technically working?
The idea shown, not described: data, layer and approach chosen. Out: a concept
Cloud demo
Does it survive contact?
Something real to click, running in the cloud. Out: a demo you can try
2–24 hoursMVP / POC
Do users adopt it?
First public customers on a real user journey, built with your team or a partner. Out: measured use
1–3 monthsFunctional product
Ready to scale?
Market-ready, scaled and operated, with someone on call. Out: a product
3–12 months+Ideas found on the way get parked. They stay in the discussion and out of the build, so the current step keeps its scope and nothing is quietly forgotten.
The full ladder. Every rung answers one question and leaves something people can look at or try. Each stage links to its own page.Source: Linda Mohamed, AI Use Cases Workshop deck, 2026
Where the facts come from.
Prices, service names and limits change. The linked official pages are the reference; figures on this page were checked in September 2026.
- AWS Well-Architected Framework
- AWS Pricing Calculator
- AWS: Amazon Bedrock pricing
- ai-solutions.wiki: From AI proof of concept to production
- ai-solutions.wiki: Total cost of ownership for AI
- ai-solutions.wiki: Build vs buy for AI
- ai-solutions.wiki: AWS funding for proofs of concept
- Regulation (EU) 2024/1689, the EU AI Act
- Documenting architecture decisions (ADR), Michael Nygard
Start with one conversation.
30 minutes to look at your idea, your data and the right starting point. Or book 15 minutes if you only have a question.
- What your demo would have to handle
- Which success criteria matter
- What data access it needs
- Whether AWS funding could cover part of it
Vienna, Austria · remote across Europe · [email protected]
