AI Copilot

AI Copilot plugs an OpenAI-compatible language model into the offboarding workflow. It doesn't replace HR judgment - it drafts, summarizes, and flags things worth a second look, and every output is meant to be reviewed before you act on it.

What it can do

AI Copilot covers six things, each scoped to a single offboarding record unless noted otherwise:

• Checklist suggestions - proposes up to 10 additional tasks for an offboarding, based on role, department, tenure, and separation type, skipping anything already on the checklist.
• Risk scoring - scores the security and access risk of a specific offboarding (0-100, with a level of low/medium/high/critical) based on open access permissions, unreturned assets, admin-level access, and how sensitive the role is. This is distinct from Flight Risk, which predicts who might resign.
• Knowledge transfer summary - turns KT notes, open checklist tasks, and exit interview responses into a structured handoff summary with suggested owners and deadlines.
• Document drafting - drafts an exit letter or separation agreement using the employee's role, department, tenure, and exit date. Only those two document types are supported today.
• Quarterly exit interview summary - an org-wide, AI-written summary of exit interview sentiment and themes for a given quarter.
• Resignation predictions - an org-wide, department-level view of which departments show the strongest resignation signals over the last 90 days, based on resignation counts and checklist SLA breach patterns.

All AI features are available to Superadmin, Admin, HR, and Manager roles, except resignation predictions, which is limited to Superadmin, Admin, and HR.

Every AI call is rate-limited to 30 requests per minute per organisation, and results are cached by input - asking for the same thing twice without anything changing returns the cached response instead of calling the model again.

Checklist suggestions and risk scoring

From an offboarding record, AI Copilot can suggest checklist items you might have missed and score how risky the offboarding is from an access and asset-recovery standpoint. Both pull from the offboarding's existing data - role, department, tenure, open access grants, and outstanding assets - so the more complete that record is, the better the suggestions.

Knowledge transfer summaries

If an employee has left knowledge transfer notes, open tasks, or exit interview responses on their offboarding record, AI Copilot can turn that into a written summary: key knowledge areas, and a list of handoff actions with suggested owners (by role, not name) and deadlines.

Drafting documents

Quarterly exit interview summary

This report aggregates exit interview responses for a given year and quarter into an executive-readable summary: an overview, key themes with sentiment, a department-by-department breakdown, and recommended actions. If there are no completed exit interview responses for that quarter, it returns a short note saying so instead of a summary - there's no minimum-response threshold beyond that.

Resignation predictions

This looks at resignation counts and checklist SLA-breach rates by department over the last 90 days and produces a short risk narrative: which departments show elevated signals, recurring patterns, early warning signs, and recommended actions. It's an organisation-wide view intended for leadership, separate from the per-employee scores in Flight Risk.

Related pages

• Flight Risk
• Exit interviews
• Integrations