Intermediate
How to configure fin ai procedures on Intercom
Quick Answer
Configure Fin AI procedures in Intercom by accessing Settings > Fin AI, creating custom procedures for specific customer scenarios, and training the AI with your knowledge base content. Set up triggers and fallback options to ensure smooth automated responses.
Prerequisites
- Admin or Owner role in Intercom workspace
- Fin AI feature enabled on your plan
- Knowledge base articles already created
- Understanding of your customer support workflows
1
Access Fin AI Settings
Navigate to your Intercom dashboard and click on Settings in the bottom left corner. Select Fin AI from the left sidebar menu. If you don't see this option, ensure your plan includes Fin AI features and contact support if needed.
Tip
Bookmark the Fin AI settings page for quick access during configuration
2
Enable Fin AI for Your Workspace
In the Fin AI settings, toggle the Enable Fin AI switch to activate the feature. Choose your preferred AI model from the dropdown menu and set the Confidence threshold slider (recommended: 70-80% for balanced accuracy). Click Save Settings to apply changes.
Tip
Start with a higher confidence threshold and adjust downward based on performance
3
Configure Knowledge Base Integration
Click on Knowledge Sources tab and select which articles Fin AI can reference. Check the boxes next to relevant Collections or individual articles. Use the Priority dropdown to set importance levels for different content sources. Enable Auto-sync to keep AI knowledge updated with article changes.
Tip
Regularly audit which articles are being used most by Fin AI to optimize your knowledge base
4
Create Custom AI Procedures
Navigate to the Procedures tab and click Add New Procedure. Enter a descriptive name and define the Trigger conditions using keywords, customer tags, or conversation context. In the Response Template field, write the AI's response format using variables like
{{customer.name}} and {{relevant_article}}.Tip
Test procedures with sample conversations before deploying to live chat
5
Set Up Escalation Rules
In the Escalation section, define when Fin AI should hand off to human agents. Set conditions such as Low confidence score, Customer frustration indicators, or Complex query detection. Choose the appropriate Team or Specific agent for handoffs and customize the transition message.
Tip
Monitor escalation rates to fine-tune your confidence thresholds and triggers
6
Configure Response Formatting
Under Response Settings, customize how Fin AI presents information. Enable Suggested articles, set Maximum response length (recommended: 200-300 words), and choose whether to include Follow-up questions. Configure the Typing indicator delay to make responses feel more natural.
Tip
Keep responses concise and actionable to improve customer satisfaction
7
Test and Monitor Performance
Use the Test Console to simulate customer queries and review AI responses. Check the Analytics tab to monitor key metrics like Resolution rate, Customer satisfaction scores, and Escalation frequency. Set up Performance alerts to notify you when metrics drop below thresholds.
Tip
Review conversation logs weekly to identify patterns and improvement opportunities
8
Deploy and Optimize
Click Deploy to Production to activate Fin AI for live conversations. Enable Learning mode to allow the AI to improve from successful interactions. Schedule regular reviews using the Performance Dashboard and adjust procedures based on customer feedback and resolution rates.
Tip
Start with a limited deployment to specific conversation types before full rollout
Troubleshooting
Fin AI is giving irrelevant responses
Review your Knowledge Sources selection and ensure only relevant articles are included. Increase the Confidence threshold and refine your procedure triggers with more specific keywords.
Too many conversations are being escalated to humans
Lower the Confidence threshold slightly and review your Escalation rules. Check if your knowledge base covers common customer questions adequately.
Fin AI is not triggering for expected conversations
Verify your Trigger conditions are not too restrictive. Check that Fin AI is enabled for the correct Conversation channels and review any conflicting automation rules.
Knowledge base articles are not being referenced correctly
Ensure articles are published and properly tagged. Check the Auto-sync status and manually refresh the knowledge sources if needed. Verify article permissions allow AI access.
Ready to get started with Intercom?
Put this tutorial into practice. Visit Intercom and follow the steps above.
Visit Intercom →