LoginSign Up
← Back to all sparks
W

WeWeb

DEVOPS
Velocity6.3

WeWeb adds LLM calls to backend workflows, so the apps it builds can now ship AI of their own

no-codeweb-app-builderllm-integrationsmcpsupabasebackend-workflows
◆Current state
WeWeb's visual builder is now working on two AI axes at once. MCP and WeWeb AI let external agents and the in-product assistant build the project, while the August 13 release lets the finished app call OpenAI, Anthropic, and Google Gemini models from backend workflows. The same release adds Make and Twilio integrations plus improved usage monitoring, and the month before it was mostly editor fixes, a refresh-token action, and load-time work.
◆Where it's heading
The build-side AI story has matured into consolidation - planning, task tracking, and workflow debugging layered onto MCP rather than new agent surfaces. The newer direction is runtime: WeWeb is becoming a place to ship AI features, not only a place where AI helps assemble a page. Supabase remains the assumed backend, and the integration list is widening toward automation and messaging rather than more databases.
◆Prediction
Expect the new backend AI actions to accumulate the plumbing a production AI feature needs - credential handling and cost controls tied to the usage monitoring shipped alongside them. The integration cadence points to more automation and messaging connectors before the next agent-side capability.

◆Recent moves

  1. 6d ago

    🤖 Add AI features to your app with new AI integrations

    ⚡ SPARK

    Until now WeWeb's AI work pointed inward - MCP and WeWeb AI helping you build the project. This release turns the arc outward: backend workflows can call OpenAI, Anthropic, and Gemini models, so AI becomes something the shipped app does rather than only something that assembles it.

  2. 6d ago

    Make and Twilio integrations, plus WeWeb AI usage monitoring

    The umbrella post for the same August 13 release: Make and Twilio join the integration list, usage monitoring improves, and WeWeb AI and MCP pick up refinements. The integration additions lean toward automation and messaging, while the AI and MCP changes continue the consolidation stretch rather than opening new agent capability.

  3. 15d ago

    Refresh token action, faster loading, and small fixes

    A maintenance release in the gap between AI feature drops - a refresh-token action, improved loading time, and small fixes. It fits the pattern of stabilizing the surfaces the AI layer drives rather than extending them.

  4. 27d ago

    Fixes across notifications, workflows, and self-hosted apps

    Cleanup across notifications, workflows, integrations, text links, and self-hosted apps. Nothing here changes the capability surface; it is the housekeeping that follows the June MCP and Supabase work.

  5. 29d ago

    📣 Improved Supabase setup + AI-assisted workflow debugging

    Role-based page access via Supabase plus AI-assisted testing and troubleshooting for backend workflows. This is the build-side AI arc deepening into debugging, and it prefigures the August release that lets those same backend workflows call models directly.

  6. 1mo ago

    🔗 Easier domain setup, cleaner publishing flows, and other improvements

    Domain connection and publishing flows get reworked, with fixes across the Editor and workflows. This is deployment plumbing rather than capability, consistent with the consolidation stretch between the June MCP release and the August AI integrations.