The useful role for AI is often smaller than “decide for me”
Managers already make decisions with incomplete information. Before a 1:1, they try to remember what matters to the person. Before changing a team, they try to understand how people work together. During hiring, they combine experience, interviews, references and other evidence.
AI can help organise that context. It can retrieve relevant information, summarise differences and turn a large amount of professional context into a useful question or consideration. That does not require the AI to decide who should be hired, promoted, managed in a particular way or assigned a particular role.
Three useful AI tasks
Retrieve the right context
Instead of asking a manager to remember every profile and report, an assistant can surface the relevant information when the manager needs it.
Make differences understandable
If two people approach decisions differently, AI can explain the difference in plain language and suggest a question the pair could clarify together.
Prepare, rather than decide
An AI assistant can help prepare a 1:1, onboarding conversation or team discussion. The user can then validate the context against reality and make their own judgement.
Transparency matters
People should know when they are interacting with AI. The EU AI Act's Article 50 transparency rules apply from 2 August 2026 and require providers of interactive AI systems to design them so individuals are explicitly informed when they are interacting directly with an AI system.
That is why HumaMap describes Ask HumaMap as an AI assistant. The interface and the marketing should not make the user guess whether the response came from a human.
Employment decisions need particular care
The AI Act also treats certain employment use cases as high-risk. Annex III specifically includes AI systems intended for recruitment or selection, including systems that analyse or filter job applications and evaluate candidates. Other work-related use cases can also be covered, including systems used for certain worker-management decisions.
The exact classification depends on the system's intended purpose and function. The Act also contains conditions under which some Annex III systems may not be considered high-risk when they do not materially influence decision outcomes, although profiling of natural persons is treated differently. This is an area where product design and legal classification need to stay aligned.
For a product like HumaMap, that means “the human clicks the final button” is not enough as a governance principle. The system should be clear about whether it is providing context, evaluating a person, ranking people or materially influencing a decision.
Context should be traceable to its source
AI language can sound more certain than the underlying information. A useful safeguard is to make the source visible. Instead of saying “Sofia needs more validation”, an assistant can say “Sofia's HumaMap indicates that she prefers more validation before committing.”
The second formulation leaves room for the user to question the profile, ask Sofia directly and recognise that a profile is context rather than ground truth.
Privacy boundaries still apply
An AI assistant should not turn access control into a suggestion. If a user cannot see a piece of information, the assistant should not use that information to answer the user's question. The same sharing model that governs the underlying profile should govern the AI layer.
For HumaMap, the intended scope is professional work context, not sensitive personal data. The product principle excludes GDPR Article 9 special category data from HumaMap profiles.
The standard should be better human judgement
The strongest reason to use AI in people work is not to automate judgement. It is to reduce the amount of context a person has to hold in their head, make differences easier to understand and help users prepare better conversations. If the AI becomes the hidden decision-maker, the product has moved into a very different category.