What is your daily routine to boost Fin AI performance?  I

​@Audrey It depends on time/effort and volume, but a general AI ops process I usually follow might look like the one below. The main goal is to create a continuous feedback loop that leads to an improvement flywheel, based on observed data.

Review

Review pending items in the Fin AI agent section within Intercom, including:

• Suggestions
• Optimize
• Topics explorer 
• Trends

Review relevant reports, such as:

• Fin AI agent report (Intercom prebuilt)

• Fin CSAT score report (custom)

• Fin unresolved conversations

• CX reports for Fin

• Other relevant reports (custom and prebuilt)

Review feedback received (if any), including:

• Manually tagged conversations with relevant tags in Intercom.

• Manually reported conversations across any communication channel 

Analysis

Analyze conversations from the reports mentioned above according to established parameters following general AI evaluation processes:
Expected outcome of a conversation vs. actual outcome → error analysis with classification → content and other improvements

Implementation (content modification or creation)
Feedback for Help Center article improvement to the relevant team (depends on who is managing this content on your end). 
Snippets
Internal articles

Continuous Feedback Loop 
Rinse and repeat; automate as much as you feel comfortable with or have the need for.