Where to start
A scored portfolio of use cases: value, feasibility, data readiness, risk and time to a first result.
Advise — AI consulting
Decide where AI pays before you pay for it.
Most AI programmes fail at the choice, not the code: the wrong use case, an unmeasured baseline, a data-protection review nobody planned. Onega’s consultants are engineers. We map your candidate use cases, measure the work they would change, classify each under the EU AI Act, design the architecture and write a business case your board, works council and data-protection officer can check.
01 — What we advise on
Six questions, answered with evidence.
What it is worth
A measured baseline and a business case with its assumptions in view, including the point below which the case does not close.
What the law requires
An EU AI Act classification per use case, the human-oversight design, and the GDPR questions to settle before any data moves.
How to build it
Architecture and build-versus-buy: which models, where they run, how they are governed and what they connect to.
Whether you are ready
Data, platform, security and skills readiness, with the gaps that would block a pilot named plainly.
What happens next
A sequenced roadmap and a fixed-price proposal for the first pilot. You may stop there.
02 — Formats
Three ways to start.
| Format | Duration | You get |
|---|---|---|
| AI Opportunity Assessment | 3–4 weeks | A ranked use-case portfolio, baselines for the leading candidates, an EU AI Act classification for each, a target architecture, a business case, a roadmap and a fixed-price pilot proposal |
| Architecture and Readiness Review | 2 weeks | A review of an AI initiative or platform you already run: architecture, security, governance, evaluation and cost, with prioritised findings |
| Discovery Sprint | 2 weeks | One workflow: map, measured baseline, data-flow map, risk register, acceptance criteria and a fixed-price pilot proposal |
03 — How it runs
An Opportunity Assessment, week by week.
- 01
Listen
Interviews with the sponsor, process owners, IT, data protection and, where relevant, the works council. A long list of candidates.
- 02
Measure
Observation and data sampling for the leading candidates: volumes, handling time, rework, exceptions and the cost of an error.
- 03
Design
EU AI Act classification, architecture options, build-versus-buy and a cost model for each shortlisted case.
- 04
Decide
Business case, roadmap and pilot proposal, presented to the sponsor for a decision.
04 — What you receive
Documents you can act on.
- A ranked use-case portfolio, with the scoring method
- Baselines measured on your work, not vendor benchmarks
- An EU AI Act classification and oversight design for each use case
- A target architecture and a build-versus-buy recommendation
- A business case with its assumptions and break-even point
- A roadmap and a fixed-price proposal for the first pilot
05 — FAQ
Questions sponsors ask.
Are you independent if you also sell products?
Every recommendation to use an Onega product states the alternative we considered and why we did not choose it, and the scoring method is in the report so you can check it. If the honest answer is that AI does not pay for a workflow, that is the answer you get.
Who do you need from our side?
A sponsor who owns the decision, the owners of the processes in scope, an IT or architecture contact, and your data-protection officer. Where AI changes how people work, we recommend involving the works council early.
What if we already have an AI strategy?
Then start with the Architecture and Readiness Review, or with a Discovery Sprint on the first use case in your plan. We test the strategy against measured baselines rather than write a new one.