Skip to main content

Which AI use cases pay back, scored against your data and the EU AI Act, in two weeks.

A fixed-fee sprint that turns a list of AI ideas into a ranked shortlist. You get each use case scored on feasibility and payback, mapped to its EU AI Act risk category, with a build plan for the ones worth doing.

At a glance

Fixed fee
€5,000
Duration
2 weeks
Format
Remote-first
Fee credit
Creditable in full within 8 weeks
Markets served
DACH and Benelux

Who it's for

  • You have more AI ideas than you can fund, and no neutral way to rank them.
  • A pilot worked in a demo and stalled on the way to production, and you want to know why before the next one.
  • You need to know which use cases trigger EU AI Act obligations before you commit a budget to them.
  • Your compliance or legal team asks about AI Act obligations before any technical conversation starts, and you need an answer that holds up.
  • You are running the sprint to build the internal case for budget, not because leadership has already agreed to fund the build.

Scope and format

Two weeks, remote-first, with one optional on-site workshop. We interview the people who own the processes, look at the data and systems each use case would touch, and assess feasibility against your Microsoft and enterprise systems landscape. Where a use case touches SAP S/4HANA, Power BI and Fabric semantic models, or your Azure AI Foundry environment, we work directly in those systems rather than from an export. The engagement covers both the AI build and the governance around it. You give us the candidate use cases and access to the people who understand them.

What you get

Ranked use-case shortlist

Each candidate scored on feasibility and payback, with the ones that will not survive your data marked as such.

EU AI Act risk classification

Every shortlisted use case mapped to its risk category, so the compliance cost is on the table before the build, not after.

Data and systems readiness read

What each use case needs from your data, your systems, and your Azure estate, and where the gaps are.

Build plan for the top use case

Scope, architecture, and the governed foundation the first use case sits on, costed enough to decide on.

Vendor and build-vs-buy read

Where a shortlisted use case is better served by an existing Microsoft or third-party product than a custom build, we say so, even when it costs us the follow-on engagement.

Stakeholder RACI for the top use case

Who owns the data, who signs off the model, and who operates it in production, agreed before the build starts rather than discovered during it.

Readout session

A working session with your stakeholders to walk the ranking and agree what gets built first.

Pricing

A fixed fee of €5,000. The scope above is the scope, with no day-rate drift.

Credit toward your build

The full fee is creditable against a follow-on build engagement booked within eight weeks of the readout.

Where your data sits

A remote-first sprint. Where we need to see data we work in your tenant with read-only access, scoped to the sprint and revoked on completion. No data leaves your environment, and we sign your data processing agreement before we start. Where data residency matters, models and processing stay in the Azure region you are bound to.

Governance & compliance

  • Each use case classified to its EU AI Act risk category as part of the deliverable.
  • Read-only access, scoped to the sprint and revoked on completion.
  • A signed data processing agreement before kickoff.
  • Feasibility and classification documented for your AI governance records.

Includes EU AI Act positioning and how we classify AI systems.

Read our governance approach

Free readiness check

The AI Readiness Check

Twenty-four questions before your first production AI workload, following the six steps of Microsoft’s Cloud Adoption Framework, with the EU AI Act obligation that applies added at each step. Work through it use case by use case, since both the Framework and the Act classify at that level, not at the level of the platform as a whole.

  • Strategy and plan: use cases classified against Article 5 and Annex III, not just identified, so a prohibited or high-risk use case is caught before the build, not after the pilot.
  • Platform and governance: Article 12 logging, a full AI inventory, Annex IV documentation as an engineering deliverable with a named owner, not a policy document nobody updates after go-live.
  • Security: prompt injection, data poisoning, and model theft in the threat model, the three attack paths a standard security review does not test for.
  • Operations: cost per use case, human oversight under Article 14, and post-market monitoring, the running cost of AI once the novelty wears off.

Verifying your browser…

Common questions

How long does it take?

Two weeks, from kickoff to the readout session.

Do you build the use case too?

The sprint ends with a build plan. The fee is creditable in full against the build engagement if you book within eight weeks. You are not obliged to.

What if none of our use cases are feasible?

Then you have saved a build budget, and the sprint says so in writing with the reasons. We would rather tell you that in week one than in month six.

Do you need access to our data?

Read-only access where a use case depends on seeing the data. For most of the sprint, access to the people who own the processes matters more than access to the systems.

Who runs the sprint?

The sprint runs as one engagement: the interviews, the feasibility read, and the EU AI Act classification feed directly into the build design if you proceed, so nothing is re-scoped from scratch.

How does the sprint's output feed into the AI Use-Case Build?

The build plan for your top use case becomes the scope document for the AI Use-Case Build: same architecture, same governed foundation, no re-discovery. Credit the fee within eight weeks and the build starts from where the sprint stopped.

What do we need to prepare before the sprint starts?

A list of candidate use cases, and two to three hours per use case from the people who own each process, spread across the two weeks. Where a use case touches a system like SAP S/4HANA or an existing Power BI and Fabric environment, we ask for read-only access to that system directly, not an export.

Do you tell us if a use case should not be built at all?

Yes, and it happens. If the payback does not clear the cost of building and running it, or an off-the-shelf product already covers it, the shortlist says so. We would rather lose that follow-on engagement than build something that should not exist.

What if leadership has not approved a budget for the build yet?

That is a normal starting point. The build plan and the EU AI Act classification are written to go into an internal budget case, with the numbers a board or Geschäftsführung would ask for.

Does the sprint cover agentic AI, or only single-model use cases?

Both. Where a use case needs multiple agents or tool calls rather than one model call, the feasibility read and the build plan account for that complexity separately.

Can the sprint run across more than one business unit?

Yes, as long as the candidate list and the interview time still fit the two weeks. Larger portfolios usually need a short scoping call first.

Request the sprint

Comparable work

Generative AI knowledge platform · RAG on Azure AI Search

Manufacturing · Chemical R&D

  • One generative AI platform serving 2,000+ R&D users across globally distributed sites.
  • Natural-language querying of chemical processes, material properties, and technologies that previously needed manual search.
All references

Verifying your browser…