Switch AI agents between business units

Workflow purpose

This guide demonstrates how to use Proto's platform to hand a customer off between two AI agents run by different business units within the same organisation – for example, from a general support agent to a specialist billing or claims agent. Large organisations often run separate AI agents per business unit for good reason – different training data, different escalation paths – but the customer shouldn't feel that seam. This workflow uses Proto's AI Agent Network to pass the conversation across with full context.

Who can benefit from this guide:

Enterprises running separate AI agents per department or business unit
Support teams handing off from general enquiries to a specialist team
Operations teams wanting a single "no wrong door" customer experience
Multi-brand organisations sharing a Proto workspace across units

Switch between business units

This workflow confirms the handoff with the customer, then transfers the conversation to the receiving business unit's AI agent with full context – unless the customer would rather stay put. See how to switch AI agents between business units.

Just starting with the platform?

For initial setup of your AI agent – including workspace setup, agent training, and channel deployment – please start here. This workflow guide assumes your AI agent is set up and ready for more advanced configurations.

1. Create a trigger

Purpose: Catch the moment a customer’s request belongs to a different business unit.

Navigate to the Actions tab in the AI agent settings.
Click “+ Add Trigger” and select the Message Received trigger type.
Name the trigger “Business unit handoff”.
Add a description such as “Triggers when the customer’s request matches a specialist team’s scope rather than general support.”

2. Confirm the handoff

Purpose: Give the customer a say before they’re transferred anywhere.

Add a Send message action: “This looks like something our {destination_team} can help with faster. Shall I transfer you? No need to repeat what you’ve told me.” with quick replies Yes, transfer me, No, keep going here.
Add a Set chat variable action to capture the reply as handoff_choice – quick reply selections aren’t captured automatically, so this step reads _user_input into the variable.

3. Branch on the customer’s choice

Purpose: Only transfer the conversation if the customer actually agreed to it.

Add a Branch action. Condition: _.lower(handoff_choice) == "yes, transfer me".
If false: Send message – “No problem, let’s keep going here.” – and end this flow.

4. AI Agent Network handoff

Purpose: Transfer the conversation with everything already collected.

Add an AI Agent Network handoff action.
Enter the destination business unit’s AI agent ID, AI agent secret, and Platform domain, then Verify & save.
Confirm the transferred context includes the fields the receiving team actually needs, e.g. {_transferred_context.issue_description} and {_transferred_context.account_reference}.

5. Handoff confirmation

Purpose: Confirm the transfer completed so the customer isn’t left wondering.

Add a Send message action: “You’re now connected with {destination_team}.”

6. Test the flow

Purpose: Confirm the receiving team gets a complete, usable handoff, and that declining actually works.

Run a test conversation, collect a few fields, trigger the handoff, and confirm the receiving agent’s test chat shows the transferred context correctly.
Test with the destination team’s agent offline or unpublished, and confirm the flow fails with a clear message rather than a silent drop.
Test declining the handoff and confirm the conversation continues with the original agent rather than transferring anyway.

550K