← HomeAsana · AI integrations
Designed Custom MCPs for Asana AI Teammates
In 2026, I designed the end-to-end Custom MCP experience for Asana AI Teammates—from developer setup and organization approval to individual consent and safe connection changes. I also prototyped the flow in code and shipped some of its UI.
- Problem
- Native integrations could not scale to every customer tool, but customer-hosted MCP apps introduced organization-wide authentication, approval, and trust decisions.
- My contribution
- Owned the end-to-end experience across Integrations, AI Teammates, AI Chat, developer tooling, and Security; prototyped interactions in code and shipped part of the UI.
- Key decision
- Used the governed Developer Console for initial setup, reused existing components, made consent transparent, and revoked connections after sensitive endpoint changes.
- Downstream effect
- Within two months, Custom MCP apps handled 30% of total API calls; the platform supported 46 unique BYO connectors in 10 weeks and 400 external domains.
This case study is private
Some details are under NDA. Enter the password or email me for access.
That password isn't right — try again.