01
Job requisitions
Open roles with headcount and status, so hiring starts from an approved position rather than a conversation.
Hire to retire · one continuous record
Most companies keep the candidate in one system, the employee in another and the leaver in a folder. Flume keeps one record and writes an event every time it changes hands.
The spine, moment by moment
This is the part that stops an HR system drifting apart from itself. Pick a moment in an employment and see which event is emitted and which modules react to it — none of which the emitting module knows about.
One employee record, five moments
The event is committed alongside the change itself, so it cannot be dropped if a subscriber is down — it is picked up when the subscriber returns.
Events are committed in the same transaction as the change they describe, so a business write cannot succeed while its announcement is lost.
What ships
01
Open roles with headcount and status, so hiring starts from an approved position rather than a conversation.
02
A drag-and-drop board through applied, screening, interview, offer, hired and rejected.
03
An external-facing application route that feeds the same pipeline, with no re-keying.
04
Scheduled interviews with scorecards, so a decision rests on recorded criteria rather than recollection.
05
Generated offer letters, signed electronically, with the signed artefact retained.
06
Type, term, probation and renewal chains — a contract knows the one it replaced.
07
Scheduled warnings before probation ends or a fixed term lapses, while there is still time to act.
08
Task checklists per new hire, assigned across the people who actually complete them.
09
Structured exit with its own task list, reason and status through to completion.
10
End-of-service, outstanding balances and deductions computed into a settlement the leaver can be shown.
11
Every joiner, mover and leaver event written durably as it happens, for anything downstream to react to.
12
Contracts, IDs and issued equipment attached to the same record, with expiry and return tracked.
How it actually works
Lifecycle events are committed alongside the business mutation they describe, using a transactional outbox rather than a best-effort notification fired afterwards. A hire cannot succeed while the event announcing it silently fails, which is the usual reason a downstream system drifts out of sync.
The notification service subscribes to the event log like any other consumer, and new consumers can be added without touching the modules that emit. That is why announcements, alerts and realtime updates stay consistent with each other instead of each module growing its own notify call.
The audit log exists to be read by people — a compliance history that never triggers anything. The lifecycle spine exists to be dispatched. Keeping them separate stops a UI history feature and an integration mechanism from being forced into one table that serves neither well.
Book a demo
The fastest way to judge Flume is to run it against a period you already argued about. Send a month of punches and the payroll you produced from it, and we will show you the same month reconciled — including what the anomaly detectors flag.