← Back to all sparks
Mixpanel logo

Mixpanel

ANALYTICSINFRA · APIS
Velocity6.7

Product analytics for user behavior insights

Mixpanel goes wide: Glean MCP, Postgres GA, AI Metric Trees — analytics is being redrawn around AI access patterns.

product-analyticsmcp-serverfeature-flagspostgres-connectorai-integrationaudit-log
Current state
April was a heavy ship month: Custom Roles for granular project permissions, a Glean integration that exposes Mixpanel as an MCP app inside Glean Assistant, Comments and a Notification Center for in-product collaboration, the Postgres Connector reaching GA so transactional data can sit alongside product events without ETL, scheduled/webhook-aware Alerts, and an Audit Log for compliance. March's spree included Mixpanel MCP for Claude/ChatGPT/Gemini/Cursor/Notion and Feature Flags expanding to Go and Flutter SDKs.
Where it's heading
Two compounding bets: turn Mixpanel into a queryable surface from any AI tool (MCP server, Glean integration, AI Metric Trees) and absorb adjacent categories (Postgres direct connection competes with reverse-ETL/CDP plumbing; Feature Flags + Experiments competes with LaunchDarkly and Statsig). Combined with collaboration (Comments) and enterprise hardening (Custom Roles, Audit Log), the product is repositioning from 'product analytics tool' to 'analytics platform that other tools — including AI assistants — read from.'
Prediction
Expect the MCP and Glean playbook to extend to more enterprise AI surfaces (Notion AI, Microsoft 365 Copilot), and the Feature Flags push to add more SDKs and likely a deeper Statsig-style experimentation analysis layer. The Postgres Connector pattern probably grows to other transactional databases like MySQL and Snowflake direct.

Recent moves

  1. 2mo ago

    Custom Roles: Granular project-level permissions

  2. 2mo ago

    Custom Roles ship with granular project permissions

    Custom Roles let admins build granular permission profiles across six categories beyond the four default roles. Up to 5 per project. Solid enterprise hardening — fits the broader push (Audit Log, scheduled Alerts) toward larger-organization deployments.

    View source ↗
  3. 2mo ago

    Mixpanel goes live as an MCP app inside Glean

    ⚡ SPARK

    Mixpanel is now an MCP app inside Glean Assistant, so users can ask Glean about conversion, retention, and feature adoption and get live Mixpanel data alongside Slack threads, Jira tickets, and Docs in the same answer. Continuation of the MCP push that started in March, but Glean specifically is the enterprise-AI route to busy product and engineering teams.

    View source ↗
  4. 2mo ago

    Comments and Notification Center for in-product collaboration

    Threaded comments on reports and boards, plus a Notification Center for mentions and replies. Standard collaboration playbook — keeps decisions tied to the data instead of scattering them across Slack and Docs.

    View source ↗
  5. 2mo ago

    Postgres Connector is now Generally Available

  6. 2mo ago

    Postgres Connector reaches GA — analyze transactional data without ETL

    ⚡ SPARK

    Postgres Connector reaches GA. Production Postgres data — user records, subscription state, AI model logs — can be analyzed alongside product events without ETL pipelines, CDPs, or custom scripts. This is the most architectural shift in the batch — Mixpanel is encroaching on the reverse-ETL and CDP plumbing layer.

    View source ↗