PlacementFlow operates at different autonomy levels depending on how much oversight you want — from approving every action to letting the AI handle routine work while you review outcomes. Autonomy is set per action type (screening, outreach, scheduling, submissions…), and you choose the level — moving up as you build confidence, down whenever you want more oversight.
| Level | Name | What It Means |
|---|---|---|
| L1 | Review Everything | Every action requires your approval |
| L2 | Review Outreach | Low-risk work is automatic, outreach needs approval |
| L3 | Supervised | Routine tasks handled automatically, edge cases flagged |
| L4 | Autonomous | Runs independently, alerts you to exceptions only |
Best for: New users, first week of operation, high-stakes roles
At L1, the AI drafts and proposes, but nothing happens without you. You approve every shortlist, every email, every submission, every reply response.
Overhead: High. Risk: Low. Nothing happens without you.
Best for: After 1-2 weeks, once you trust the sourcing
At L2, the AI sources and ranks candidates automatically and handles low-risk work, but queues emails and outreach for your review. You approve outreach content, submissions to clients, and anything unusual.
Overhead: Medium. Risk: Low. Outreach still needs your sign-off.
Best for: Established workflows, trusted briefs, experienced users
At L3, routine tasks run automatically — standard reply handling (Interested, Not Interested, Out of Office), follow-ups, scheduling coordination. Edge cases are flagged for you: counter-offers (always), unusual reply classifications, client escalations, offer negotiations.
Overhead: Low. You're handling exceptions only. Risk: Medium. The AI is acting on your behalf.
Best for: High-volume desks, proven track record, maximum throughput
At L4, the AI runs the loop independently and escalates only hard exceptions. You review outcomes on the dashboard and intervene when metrics drift.
Note: L4 never enables itself. You must explicitly set it — there is no automatic promotion to autonomous operation.
Autonomy isn't one global dial. Each action type has its own level, and some are capped for safety:
Autonomy levels are yours to set — the AI never promotes or demotes itself. What the system does give you is a complete audit trail: every AI-driven action, and every time you approve or reject one, is recorded. That log is what lets you decide, per action type, whether you're comfortable moving a level up or want to pull one back.
A practical way to build confidence: run an action type at L1/L2 for a week or two, review how the AI's drafts and classifications look against your judgement, and raise the level once you trust it. Counter-offers always route to a human regardless of level (see below).
Changes take effect immediately for new actions.
Each AI desk can override the agency-wide settings — useful when one desk runs high-stakes retained searches (more oversight) and another runs high-volume roles (more autonomy). Manage overrides from the desk's autonomy settings.
| Situation | Recommended Level |
|---|---|
| First week | L1 (Review Everything) |
| Weeks 2-4 | L2 (Review Outreach) |
| Established workflow | L3 (Supervised) |
| Proven performance, your explicit call | L4 (Autonomous) |
Start conservative and raise each action type as you build confidence in it.