Activating & Deactivating Agents
Understanding Agent Active Status
Each Agent Definition has an Active Status checkbox that controls whether the agent is available to respond to user messages. This setting is fundamental to agent management:
- Active Status checked — The agent is live and responds to user messages immediately.
- Active Status unchecked — The agent is offline. Users cannot interact with it; they see a "not available" message instead.
The active status applies globally — if an agent is inactive, it's unavailable on all deployment channels (Internal, External Widget, Community, API).
What Happens When an Agent is Inactive?
| Scenario | Behavior |
|---|---|
| User opens chat widget | Chat window appears, but a "not available" message is shown instead of the chat input. User cannot send messages. |
| User tries to send a message via API | The API call returns an error status indicating the agent is not active. |
| Flow invokes the agent | The Flow receives an error response. The agent does not process the request. |
| Existing conversations | All conversation history is preserved. Reactivating the agent does not delete any data. |
Why Deactivate an Agent?
There are several valid reasons to temporarily deactivate an agent:
- Maintenance — Pause the agent while you update prompts, tools, or configuration.
- Testing — Test changes in sandbox before bringing the agent live again.
- Plan Limits — Your REDA AI plan may limit the number of active agents. Deactivate unused agents to stay within the limit.
- Seasonal Use — Deactivate agents that are only needed during certain times of year (e.g., recruiting agent in hiring season).
- Version Updates — Deactivate the old version of an agent while deploying a new version.
- Troubleshooting — If an agent is misbehaving, deactivate it temporarily while you investigate.
Active Agent Limits by Plan
Your REDA AI licence determines how many agents can be active simultaneously. The free plan allows 1 active agent; paid plans carry their own limit in the licence key issued by REDA. Check your plan details in the Admin tab to see your limit.
If you try to activate an agent beyond your limit, the save is blocked with the message "Cannot activate this agent. Your plan allows a maximum of X active agent(s). Please deactivate an existing agent first or upgrade your plan." Deactivate another agent, or contact the REDA team at sales@reda.one to raise your limit.
How to Activate an Agent
-
Open the Agent Definition Record
In the REDA AI app, go to Agent Management and click on the agent you want to activate.
-
Locate the Active Status Field
Scroll down the Agent Definition record to find the "Is Active" checkbox.
-
Check the Checkbox
Click the checkbox to mark it as checked.
-
Click Save
Save the record. The agent is now live and will respond to user messages.
-
Verify Activation
Open the chat widget on a page where the agent is deployed and confirm you can interact with it.
How to Deactivate an Agent
-
Open the Agent Definition Record
In the REDA AI app, go to Agent Management and click on the agent you want to deactivate.
-
Locate the Active Status Field
Scroll down to find the "Is Active" checkbox.
-
Uncheck the Checkbox
Click the checkbox to unmark it.
-
Click Save
Save the record. The agent is now offline.
-
Verify Deactivation
Open the chat widget where the agent was deployed. You should see the "not available" message instead of the chat input.
Best Practices
Always deactivate agents rather than deleting them. Salesforce will refuse to delete an Agent Definition while it still has Custom Prompts, Context Fields, or Suggestion Chips attached — the delete fails with an error — and conversations are never deleted with an agent: they remain in your org, detached from it. Deactivation is the clean path: it keeps the entire configuration intact and reversible.
Monitor Active Agent Limits — Keep track of how many agents are active against your plan limit. Set up a regular review (quarterly) to deactivate agents no longer in use and free up slots for new agents.
Document Why an Agent is Inactive — If you deactivate an agent, record why and when it might be reactivated — in a custom field your team adds, or in your team's own documentation (the Agent Definition has no built-in description field). This helps future admins understand the agent's status.
Test Before Full Activation — When activating a newly configured agent for the first time, test it in a sandbox environment first. Once satisfied, activate it in production.
Data Preservation During Deactivation
Deactivating an agent does not delete any data:
- Conversation History — All past conversations are preserved in the agent's conversation records.
- Configuration — Prompts, tools, context fields, and settings remain unchanged.
- Related Records — All related records (Custom Prompts, Context Fields, etc.) remain intact.
When you reactivate the agent, it starts fresh with the current configuration, but all historical data remains available for reporting and audit purposes.
Activation & Deactivation Audit Trail
REDA AI ships the Agent Definition object with field history tracking already enabled, so you do not have to switch anything on. Is Active is one of the 28 tracked fields, alongside the ones that change an agent’s behaviour or reach — Persona, Channel, Deep Reasoning, Execution Context, Allowed Origins, Economy Tier Consent, the rate limits and the External API Key. Open the Agent Definition record and look at its history related list to see who flipped a setting and when.
Note this is record history, not the Setup Audit Trail — the Setup Audit Trail logs configuration and metadata changes, not edits to Agent Definition records. Field history is the one that answers “who deactivated this agent?”
Bulk Activation/Deactivation
If you need to activate or deactivate multiple agents at once:
- Use Data Loader to export all Agent Definition records.
- Modify the Active Status column (true or false) in the CSV file.
- Use Data Loader to update the records with the new active status.
Frequently Asked Questions
Yes, absolutely. Deactivating an agent simply turns off its Active Status. You can reactivate it anytime by checking the checkbox again. All configuration and conversation history are preserved.
Existing conversations continue to work until the session times out (typically 30 minutes). New messages from new users will receive the "not available" message. It's best practice to inform users that an agent will be deactivated during a maintenance window.
No. If you try to activate an agent and you're already at your limit, Salesforce will show an error. You must deactivate another agent first, or upgrade your plan to a higher tier.
Yes. If a Flow or external system is calling the agent via API, the call will fail with an error indicating the agent is not active. You must reactivate the agent or update the integration to use a different agent.
Next Steps
Learn how to optimize agent configuration with Build Agent with AI, or explore prompt management to fine-tune your agent's responses.