Human approval is mandatory
The line manager remains responsible for accuracy, fairness, relevance, and the final wording.
AI-assisted employee onboarding
Use AI to draft clearer onboarding outcomes. Keep the manager accountable.
The Primi Centum SMART assistant helps managers turn role aims and daily expectations into draft outcomes, pass standards, and evidence requirements. Every suggestion remains editable and requires human approval before it becomes part of a programme.
Human-controlled authoring
A bounded drafting workflow with a manual route.
This public description explains responsibility and product behaviour. It intentionally excludes confidential implementation details and customer configuration.
Provide context
The manager supplies the approved role aim, operating context, and expected work.
Receive a draft
The assistant proposes a concise outcome, pass standard, and evidence requirement.
Check the wording
The manager reviews accuracy, relevance, proportionality, and fairness.
Edit and approve
The manager changes the suggestion and approves only the final authorised wording.
Publish the expectation
The new hire sees the approved outcome, not an unreviewed AI draft.
Illustrative example
From a broad aim to a checkable expectation.
The example is fictional and simplified. It demonstrates the type of drafting assistance, not confidential implementation detail or a customer result.
- Manager input
- Grow the company presence across LinkedIn, Instagram, and Facebook. Establish a consistent publishing rhythm and help recruiters contribute.
- Draft outcome
- By Day 30, maintain a four-week approved content calendar across the agreed channels, with scheduled publishing responsibilities and a weekly performance review.
- Draft pass standard
- Each planned item has a channel, owner, and due date; required approval is recorded; missed activity has an assigned follow-up action.
- Draft evidence
- Approved calendar, published links, approval record, and weekly performance summary.
Responsible boundary
Drafting assistance, not automated judgement.
No autonomous readiness decision
The assistant does not decide hiring, dismissal, promotion, probation, suitability, or performance.
Manual authoring remains available
A manager can create and edit outcomes without using generative AI.
Data handling
The assistant uses the role and programme authoring context needed to prepare a suggestion. Managers should not enter special category data, passwords, unrelated employee information, or confidential material. Provider, processing, retention, and contractual terms are confirmed in the applicable customer documentation.
External governance context
Meaningful human oversight matters.
The UK Information Commissioner's Office explains that a human reviewer should remain engaged, critical, and able to challenge an AI-supported recommendation. Primi Centum applies that principle by limiting AI to editable authoring support.
Questions and answers
AI-assisted onboarding FAQs
Does the Primi Centum SMART assistant make employment decisions?
No. It drafts authoring suggestions. An authorised human reviews, edits, approves, and owns every final outcome and employment-related decision.
Can a manager edit the AI suggestion?
Yes. The suggestion is editable and should be changed whenever it is inaccurate, disproportionate, unclear, or unsuitable for the role.
Is there a manual alternative?
Yes. Managers can author SMART outcomes without using the assistant.
What information should managers avoid entering?
Managers should not place special category data, passwords, unrelated employee data, or confidential material into authoring text.
See AI-assisted SMART drafting with human approval.
The walkthrough demonstrates the manager experience without disclosing confidential implementation details or customer content.