Skip to main content
Who can use this?
  • Owners and Admins. Custom roles with Can build and deploy Agents or Can manage Agents.
  • Needs the Advanced plan. On Basic and Essential the Post-session settings button is locked.
After a session ends, Gallabox can pull details out of the chat, score it, and send the results to your system. You open this from Analyze, next to Export. It is not in Settings.
Analysis starts when the session is Ended or Dropped. It does not run while the session is still Active. On Advanced, every ended or dropped session gets an AI summary and a Success evaluation against the agent’s Goal on Home. Extra fields, custom scores, and a webhook need the settings below.

​
Plan Availability

On a lower plan the button shows: Post-session settings are not available for your plan. Please upgrade.

​
Open Post-session settings

  1. Go to AI Agents and Bots.
  2. Open the Chat AI Agent.
  3. Click Analyze.
  4. Click Post-session settings (next to Export).
The panel title is Post-session settings. Changes Auto saves.
If you do not see Analyze, the agent is on the older OpenAI Assistant setup. Current Chat AI Agents show Analyze.

​
Data extraction

Automatically pull structured fields from the conversation when the session ends. This uses AI credits. Click Add. Fill:
  • Data Type: Boolean (True / False), String (Text response), or Number
  • Name (required). Example: customer_email
  • Description (required). Tell Gallabox what to look for. Example: Extract email address mentioned by the user
For String only, you can add Allowed Values (Optional) to limit the result to a set list. The screen shows how many fields you have added. Test Extraction is available after at least one session exists. With no sessions, it stays disabled: No sessions available to test.

​
Outcome evaluation

Add extra custom scores for this agent. This uses AI credits. Successful sessions on Analyze still uses the agent’s Goal on Home. Click Add. Choose Pick from template or Create new outcome evaluation. Fill Name and Description. Types match data extraction: Boolean, String, or Number. For String only, you can add Allowed Values (Optional). The screen shows how many criteria you have added. Test Evaluation also needs at least one existing session. Edit a criterion from the list.

​
Webhook setup

Send the extracted fields and evaluations to your own URL when a session ends. You set it up on the agent. It also appears in Settings → Webhooks as Agent Analysis. Click Add. Fill:
  • Webhook URL (required)
  • Description (optional)
  • Secret (optional)
  • Headers (optional)
Gallabox sends the event Conversation.AI.Analyse. You can have one of these webhooks on the agent. Edit or delete it from the same panel.

​
What you see after a session ends

Results live on the session’s Analyze tab. See AI Agent Analytics. If the drawer says This session is being analysed, wait, then reopen the session. A session with no messages is skipped.

​
How this differs from drop-off

Drop-off can end the session. Post-session settings then analyze that ended session. Full drop-off detail: Channel Behaviour.

​
FAQs

Your plan is not Advanced. Upgrade, or keep using Analyze for Total, Ended, and Dropped sessions. Those cards stay available.
No. Meta’s 24-hour window is when you can send free-form WhatsApp or Instagram messages. See Channel Behaviour.
You need at least one session for that agent, and that session needs at least one message. Send a test chat, wait until the session ends, then try again.
Yes. Data extraction and Outcome evaluation both say they use AI credits. See AI Wallet.
The same analysis can run when a Voice AI Agent session ends. Set it up from that agent’s Analyze tab the same way. Voice AI Agent is covered under WhatsApp Calling.