Take off without AI.
Go further with AI.
Every workflow in HR-Autopilot runs on fixed rules, with no artificial intelligence at all. Where it pays off, you add an agent. Per workflow, revocable at any time, always visible in the log.
AI is a switch,
not a requirement.
Many vendors sell artificial intelligence as the reason to buy their product. For us it is one tool among several. It only works well when the foundation is right, and we build that first.
If this, then that. Every step is set, every run identical, every result predictable. That is enough for most workflows.
- Fully traceable
- No model, no probability
- Reviewable by the works council and internal audit
An agent reads what no rule can capture: free text, certificates, CVs. It prepares and suggests, and people still decide.
- Understands informal input
- Spots anomalies in figures
- Can be switched off individually at any time
Where AI helps
and where it does not.
The same workflow, once with a fixed rule and once with an agent. The last column says whether the difference is worth it.
Fixed rule: create the record, set up the accounts, send the messages. Identical every time, traceable every time.
The agent reads the signed contract, picks out the start date, the role and the site and fills in the fields itself.
No advantage from AI when the data is structured anyway.
Fixed rule: record the absence, watch the deadline, notify the manager.
The agent reads the informal email or the photo of the certificate and enters the period and the type itself.
A clear advantage, because sick notes rarely arrive in a structured form.
Fixed rule: acknowledge receipt, sort by role, set a deadline.
The agent summarises the CV, compares it with the requirements and suggests a rating. A person makes the decision.
A big time saver with high volumes, but never without human approval.
Fixed rule: send the request to whoever is responsible, then write to both systems once approved.
The agent checks whether the change fits the collective agreement and the budget, and points out any discrepancies.
Useful with complex collective agreements, otherwise the fixed rule is enough.
Fixed rule: fetch the metrics from every system, put the report together, send it.
The agent puts the anomalies into words: what has changed and what might be behind it.
The report gets produced either way. AI turns it into an interpretation.
Six promises
we hold to.
The foundation first.
Then the intelligence.
In the call we build a workflow with a fixed rule and then add an agent. You see the difference on your own data.
