How you control the workforce

Autonoms is controlled through a single command layer: the orchestrator plus direct chats with agents.

The orchestrator

Use the orchestrator when you want to give outcomes, not tasks.

Good orchestrator prompts:

• “Book 40 qualified meetings with US companies hiring engineers.”
• “Focus on US SaaS companies with 11–50 employees this week.”
• “Increase follow-up cadence for warm prospects and report EOD.”

The orchestrator will route work to the right specialists and keep the workflow coordinated.

Talking to individual agents

Use direct chats when you need a specialist action:

• Research: “Do these accounts fit our ICP? What’s their trigger?”
• Outreach: “Draft a variant emphasizing security; keep it <60 words.”
• Qualification: “Raise the threshold; only promote score ≥80.”

Human Operator messages

Autonoms supports a human-in-the-loop model. Human Operator messages are distinct and auditable.

Use Human Operator involvement for:

• approvals (templates and campaigns)
• compliance-sensitive language
• ambiguous qualification decisions
• escalation and exception handling

Best practices

• Be explicit about constraints (regions, excludes, daily limits, compliance requirements).
• Ask for a summary at a cadence (daily/weekly).
• Treat the inbox as the source of truth for decisions (“what did we decide and why?”).