Context

Document how each driver team has implemented their AI/ML integrations. Note any similarities/commonalities in issues across all implementations.

Key points to capture:

• Type of MongoDB server used for testing (local/cloud)
• Type of LLM used for testing (Homemade Mock Embedder / Paid for OPENAI_API_KEY)
• How often regression are caught
• All the libraries our teams have integrated with within a driver team
• How often tests get run within the integrated repository (and outside of it)
• Where and if there are any "implementation best practices" for their team
• Findings on how requests surface
• Findings on how integrations get prioritized/triaged
• Findings on how teams deal with external PRs
• ...etc

Definition of done

Documentation of investigation findings. Potential follow-ups.

Pitfalls

What should the implementer watch out for? What are the risks?