Bot autonomy modes and human approval (HITL)

Overview

Autonomy defines how much the Bot (Maestro acting inside the conversation) does on its own versus when it asks a person to confirm. You pick the level of trust, and it governs how the Bot behaves in support.

There are three modes:

• Autopilot — the Bot executes on its own, without waiting for approval.
• Copilot — the Bot proposes cards (a message draft or a module action); you review and send. Nothing goes out on its own.
• Hybrid (Require approval / HITL) — the Bot acts on its own in day-to-day work, except for sensitive actions, which are held, waiting for your approval.

The safety principle is simple: in Hybrid, nothing sensitive or final happens without approval; in Copilot, nothing is sent without you. You decide how much to trust.

This concept has analogs elsewhere in the platform: the Brain's departments have their own autonomy level (including an "Ask first" tier), and Follow-ups have AI-draft approval. Both are linked under "See also".

Prerequisites

• Maestro enabled and a Bot/agent configured on the inbox.
• Administrator permission to set the autonomy mode.
• An account administrator to approve: the approval notification, the shield and the queue reach administrators only. The conversation's assignee (or, failing that, its team) is @mentioned in the private note.
• The per-module tools you want proposed/parked must be enabled — see the tools article.

Step by step

1. In the Bot configuration, choose the autonomy mode: Autopilot, Copilot or Hybrid (Require approval).
2. To change only one conversation, open the Maestro side panel, click the Autonomy mode card and choose another mode. The card says Set for this conversation; use Use Bot default to remove the exception. Only administrators can make this change.
3. In Copilot: the Bot proposes cards (a draft or a module action); you review and send/confirm — nothing goes out on its own (details in the generative copilot article).
4. In Hybrid: for day-to-day actions the Bot acts; when a sensitive action comes up, it is parked — the Bot posts a private note that @mentions the conversation's assignee or, failing that, its team (the mention fires the native notification) and creates an approval card with an action id.
5. An account administrator taps/clicks the Maestro indicator in the conversation header to open the panel; the adjacent shortcuts Pause, Resume, or Take over the service. Pending approvals stay visible in the numbered shield. In the panel, choose Approve or Reject. On Approve, the exact action that was held is re-executed; the note records who approved it. Only account administrators see the numbered shield and the approval queue, and only they can Approve or Reject. Other agents see the private note in the conversation, but not the queue.
6. In Autopilot: the Bot executes on its own (sensitive actions are still recorded in the audit trail).
7. Track the result on the card and in the conversation's activity line.

Settings & options

The three autonomy modes

What counts as a sensitive action in Hybrid (gets parked)

• Assign to a team, run a macro, mark the conversation as resolved.
• Writes to CRM, Tasks, Calendar (create/edit/reschedule/cancel an appointment) and Payments.
• Commerce recovery (a recovery card sent to the customer), cancel a follow-up cadence, enroll into a distribution group and trigger a collaboration round.
• Database writes, generate video, contracts (create/send/remind) and delegating to a sub-agent when that delegation is marked as sensitive.

What the Bot does freely (even in Hybrid)

• Reply to the contact, add labels, internal notes, reactions, generate an image, knowledge-base searches and other read-only lookups.

Contracts sent by the Robot

Sending a contract for signature is a sensitive action: in Hybrid mode it is parked for approval. In Autonomous mode, what the Robot may do is decided by the account in Contracts > Settings > AI agent, with both options off by default:

• The AI agent can send contracts for signature: off, the Robot does not send and reports that a person must send. On, it sends only to the contact of the conversation it is handling.
• The AI agent can sign for the company automatically: off, a contract sent by the Robot waits for a person to click Sign for the company on the contract page. On, the AI agent signs legally binding documents on behalf of the company, with no person reviewing them.

If automatic company signing is disabled before the signature occurs, the sent contract waits for the company; signatures already applied remain valid. An action a person approves in Hybrid mode runs with that person's permissions and follows the human send rule. If the Message is recorded in the conversation but its publication is pending, the result reports publication_pending: check the delivery before retrying and retain the same attempt key.

The Send contract for signature tool must also be enabled in the Robot's tools. Enabling it there does not replace the Contracts options: the platform checks both on every send.

Human-only action

• Approving an AI Follow-up draft is always human — the Bot never approves its own draft. This holds in any mode (see the Follow-ups article).

The approval card and note

• The private note @mentions the conversation's assignee (or, failing that, its team) and references the action id; the mention triggers the platform's native notification for whoever was mentioned.
• The approval card sits in the conversation's Maestro panel, with Approve and Reject, and is shown to account administrators only.
• The decision is recorded with who approved — and on approval, the exact parked action is re-executed.

An approved macro runs with the approver's permissions

• When the Bot parks run a macro and an administrator approves it, the macro runs as the approver: with that person's permissions, conversation access and identity, not the Maestro technical user's. Messages the macro sends come from the approver, and an assign-to-self action assigns that person.
• Before approving, the platform checks that you can run that macro on that conversation. If you can't (for example, someone else's personal macro, or one published to another audience), the approval is refused with the reason, nothing runs, and the action stays pending for someone who is allowed, or to be rejected.
• Permission is checked again when the macro runs: if it is removed between approval and execution, the macro does not run.
• When the Bot runs a macro on its own, in Autopilot, it still uses the Maestro technical user.
• Flow Builder human approval: a flow that uses the Human approval step to approve a Bot's parked action no longer approves a macro. The step ends with a failure and the macro stays pending, without running, because it must run with a person's permissions and the flow has none. To migrate, leave the macro for an administrator to approve or reject in the Maestro panel and adjust the flow to handle that step's failure (for example, by notifying the team), without repeating the same approval automatically. Other actions can still be approved by the flow.

Handoff to a human: status, priority and summary

When handing a conversation off to a human team, the Bot sets the conversation's status (open, pending, snoozed or resolved) and priority (none, low, medium, high or urgent). Each one can be a fixed value or "Maestro decides": in that mode the Bot itself picks the most suitable value per conversation, at the moment of handoff — a fixed value always prevails over the Bot's choice. The handoff note sent to the agent includes the reason and a conversation summary (generated on the spot when there is no accumulated summary yet), so the person can take over with context.

Taken over by a human (the bot stands down, then comes back)

When a person replies in the conversation (from the dashboard or the mobile app) or pauses the bot, it stands down on that conversation — it stops replying so it does not talk over the human. This applies to that conversation only; the bot keeps working everywhere else.

Two options control the comeback:

• With auto-resume off, the bot stays paused on that conversation until a person hands it back — there is no time-based return.
• The clock runs from the last human message: if the person replies again, the window restarts.
• The minimum accepted window is 1 minute; anything shorter is raised to that floor.
• The setting is per bot (it applies to every inbox the bot answers on).

Where the mode applies, and the per-department analog

• The mode configured on the Bot is the default for every inbox it serves. A conversation may keep a persistent exception; precedence is conversation → Bot. The change starts on the next turn, does not interrupt an in-progress reply, and neither executes nor discards existing approvals.
• The Brain's departments have their own autonomy level (including "Ask first") — a separate setting, detailed in the departments article.

Use cases

• A team new to AI: start on Copilot and review everything before sending.
• A trusted operation with guardrails: Hybrid — routine replies on their own, money/data moves waiting for approval.
• High volume, low risk: Autopilot for simple touches, or Hybrid for most cases.
• Sensitive sub-agent delegation kept under approval in Hybrid.

Tips, limits & best practices

• Approval requests describe the action in plain language and show team or person names when they can be confirmed within the account. If a name is unavailable, verify the destination before deciding; a numeric reference does not confirm identity.

• If the queue does not load, use Try again. A loading error does not mean there are no pending requests.

• Grow trust gradually: Copilot → Hybrid → Autopilot.

• Make sure an administrator watches the approvals: only administrators receive the approval notification and can decide. The assignee (or the team) is only @mentioned in the private note.

• Review pending items often so the flow doesn't pile up.

• Approving re-executes the exact parked action — check the data before approving; if something changed, reject and ask again.

• Only enabled tools are proposed; keep the more powerful ones starting with approval.

Troubleshooting

• "I can't approve a macro": the macro runs with your permissions. Check that you can run it (personal and selective macros have a restricted audience) and that you can access the conversation. Someone who is allowed can approve it, or you can reject the action.
• "My flow stopped approving a macro": the Flow Builder Human approval step does not approve macros, because they run with the approver's permissions. Approve or reject the macro in the Maestro panel and adjust the flow to handle the step's failure, without repeating it automatically.
• "The action didn't run": in Hybrid, a sensitive action is held — an administrator has to approve it in the Maestro panel.
• "Nobody saw the approval request": the approval notification goes to the account's administrators — check their notifications. The private note @mentions the conversation's assignee or, failing that, its team.
• "The approvals queue shows an error instead of the list": Maestro could not read the queue at that moment. Nothing was approved, rejected or removed, and approvals may still be waiting — that is why the queue is not shown as empty. Use Try again in a moment; meanwhile, the private notes in the conversations still show what is on hold.
• "The Bot acted on its own on something I wanted to review": it's not on the sensitive list, or the mode is Autopilot — switch to Hybrid.
• "I can't approve an AI Follow-up draft through the Bot": draft approval is human by design — see the Follow-ups article.
• "Approve doesn't work and the card says Maestro is paused": after a handoff to a human, or when someone pauses or takes over the conversation, the Bot stands down in that conversation and no approval can run there. Nothing was run. If the action is no longer needed (for example, the transfer already happened), reject the proposal. If it still is, resume Maestro in the conversation, then approve. A team transfer proposal is closed on its own when the Bot itself already handed the conversation off: it leaves the panel, and a click on an outdated screen says it was superseded by the handoff the agent made, without running anything.

Review an approval

Text follows the interface language without changing registered names or execution parameters.

After you approve or reject, the interface confirms, in the language you use it in, when the decision was applied. When it was not applied, it explains what happened to the attempt: nothing ran and the proposal is still pending (resolve the cause, for example by resuming Maestro, and approve again); nothing ran and this proposal can no longer run (reject it and ask for a new action); the proposal is no longer pending and nothing on record says why, so it may have been approved in another tab (check the conversation history before asking for a new action); the attempt was used with no effect (it will not run again; ask for a new action); or the result was not confirmed (check the conversation before any new action and do not approve again).

Decision buttons are unavailable during refresh or execution. A queue loading failure offers Try again, which does not approve anything. When delivery needs verification, check the outcome before proposing another action; do not repeat a transfer just because confirmation was delayed.

See also

• What Maestro AI and the Account Brain are
• Generative copilot: propose, confirm and execute
• Departments, risk analysis and insights
• Maestro tools by module
• AI with human approval (human-in-the-loop) in Follow-ups