Hire an AI consultant on evidence, not on a demo.
A convincing pitch is easy. Designing and deploying a safe, affordable AI solution is something else. Our assessment tests AI consultants, AI engineers and solution architects on what they can really do, before you sign the contract.
Who is it for?
- Companies hiring AI expertise: Select freelancers, agencies and implementation partners on skill rather than on their sales pitch.
- Recruiting AI specialists: Assess candidates for AI engineer, automation lead or AI architect with the same tasks and the same rubric.
- Consultants who want to stand out: Show what you can do with an independent report, rather than only describing what you have done.
What we assess
- From question to solution: Finding the real problem behind the request, and saying so when AI is not the best choice.
- Architecture: Workflow or agent, RAG or fine-tuning, which systems and which integrations.
- Model choice: Weighing quality, cost, speed and privacy per step, with arguments.
- Data and RAG: Sources, freshness and permissions, so nobody sees what they should not.
- Security: Prompt injection, least privilege and human approval for irreversible actions.
- Evaluation: A test set, measurable criteria and regression tests before and after every change.
- Cost and scale: A realistic monthly cost estimate, with ceilings and fallbacks.
- Governance: GDPR, the EU AI Act, documentation and who is responsible for what.
What the assessment looks like
- Knowledge and decision scenarios: Randomised questions and scenarios with several defensible options, from a secure item bank.
- Architecture case: Working a realistic business question into a design, risks, cost estimate and implementation plan.
- Practical lab: Diagnosing and improving a failing AI workflow in a sandboxed environment. AI use is allowed.
- Verification interview: A structured conversation in which the consultant explains and defends their own choices. This shows who truly understands them.
An AI assistant for recruiters
An organisation with 40 recruiters wants an assistant that drafts job ads, answers candidate questions and summarises CVs. There are 3,000 internal documents, some of them confidential.
- Which parts become a fixed workflow and which an agent?
- How do you prevent confidential documents from reaching the wrong user?
- How do you measure before go-live whether it works well enough?
- What does it cost per month at 2,000 conversations, and where is the ceiling?
- Where must a person keep making the decision, and why?
Frequently asked questions
May the consultant use AI during the assessment?
Yes, in the case and the lab. A good consultant uses AI to work faster. We score what is delivered and whether the output is checked. The verification interview shows whether the choices are really understood.
How long does it take?
About 2.5 hours, split into an online part and a 30 minute interview. The online part can be done in one or two sessions.
Can we use it for several candidates or vendors?
Yes. Each candidate gets variants of the same tasks and is scored with the same rubric, so you can compare them fairly.
Can we use a case from our own organisation?
Yes, for a tailored version we can base the architecture case on your own situation, without confidential data in the task.
What does it cost?
During the pilot we agree a price per organisation, depending on the number of candidates and whether you want a tailored case. Get in touch for a proposal.
Further reading