Prove AI skills. Don't just claim them.
Everyone says they can work with AI these days. Our assessment shows who truly understands AI, applies it effectively, builds reliable agents and manages the risks. For employees, candidates and AI consultants.
Everyone says they know AI. How can you be sure?
- Claims are cheap: "Experienced with ChatGPT and AI agents" says nothing about what someone does when an API fails, a source is wrong or customer data comes into play.
- A quiz does not measure skill: Anyone who can look up terms passes a knowledge test. Real skill shows in decisions and in what someone delivers with AI.
- Mistakes get expensive: An agent that invents a status, a chatbot that quotes confidential documents or a workflow without a cost ceiling: these are not theoretical risks.
Three assessments, one free check
- AI Skills & Literacy: Employees, managers, graduates and all AI users. Measures whether someone uses AI sensibly and productively in daily work.
- AI Agents & Automation: Developers, operations and automation specialists. Tests whether someone can build, debug and secure reliable workflows and agents.
- AI Consultant & Architect: AI consultants, AI engineers and solution architects. Shows whether someone can design, justify and safely deploy a complete AI solution.
12 domains that together make up AI craftsmanship
- AI fundamentals: Language models, tokens, context windows and why output varies
- Comparing models: Weighing quality, cost, speed and privacy per task
- Prompt engineering: Instructions, schemas, structured output and testing
- Critical evaluation: Spotting hallucinations, verifying sources and figures
- RAG and knowledge bases: Embeddings, retrieval, chunking and permission filters
- AI agents: Agent loops, tool calling, handoffs and human oversight
- Integrations: APIs, webhooks, MCP, databases and authentication
- Security: Prompt injection, least privilege and data protection
- Evaluation and debugging: Test sets, tracing, regressions and release gates
- Performance and cost: Token costs, caching, latency and fallbacks
- Governance: GDPR, EU AI Act, documentation and fairness
- Business and architecture: Process analysis, ROI, system design and implementation
How the assessment works
- Choose a profile: AI user, developer, automation specialist or consultant. The profile sets the domains and level.
- Knowledge questions: Randomised questions from a secure item bank, with variants per session.
- Decision scenarios: Realistic scenarios with several defensible options. You choose and weigh up.
- Practical lab: A real task in a sandboxed environment. We score the result and the approach.
- Verification interview: For consultants and architects: a short, structured conversation about the chosen solution.
- Report: Skills profile per domain, strengths, development areas and concrete next steps.
Fix a failing support agent
A customer asks about the status of an order. The order API keeps returning an error, but the agent keeps retrying and finally reports that the order is confirmed. Locate the fault, design a better workflow with a retry policy and a safe fallback, and write test cases that prevent this from happening again.
Claimed versus demonstrated skill
We compare what someone says they can do with what they show in the tasks. A large gap is not proof of bluffing: self-assessment, test conditions and measurement error play a role too. It is a clear starting point for the follow-up conversation.
Frequently asked questions
What does the AI Skills Assessment measure?
Whether someone understands AI and applies it sensibly: from prompting, model choice and verification to RAG, agents, integrations, security, evaluation, cost and governance. Per profile we select the domains that matter for the role.
Can candidates use AI during the assessment?
In the practical tasks, yes. We then score the result and the approach: what someone checks, which choices they make and how they handle errors and risks. In the knowledge questions, no.
How is this different from an AI quiz?
A quiz mainly measures whether you know terms. This assessment combines knowledge with decision scenarios, a practical lab and, for specialists, a verification interview. That shows the difference between knowing and doing.
Is the assessment validated?
It is currently in a pilot phase. We are collecting pilot data to support reliability, fairness and norms. Until then, use the outcome as input for a conversation, not as a hard selection threshold.
Can we use it for AI literacy under the AI Act?
Yes, as a baseline per role. Article 4 asks organisations to take measures that support AI literacy; no specific level or certificate is required. A measurement shows where training pays off most and helps you document your approach.
Which languages are available?
Dutch, English, German and Spanish. Tasks and scoring criteria are equivalent in content; norms are compared per language separately.
Can we use the assessment in our own system?
Yes. We can integrate it into your careers site, ATS or an own hiring platform we build for you, with your branding and with the data owned by you.
Further reading