AutoKeren — AI Application Deployer with GitLab MCP

AutoKeren Developers

What it does

AutoKeren is an AI-native application deployer. Users describe any AI product in plain language — AutoKeren's multi-agent engine builds the full stack (UI, backend, Cloud SQL database, Cloud Storage, auth) and deploys it to a public URL automatically.

As a live demonstration, we built BizBrain AI entirely through AutoKeren: an autonomous business operations agent that ingests documents, reasons with Gemini, and executes actions via Model Context Protocol (MCP) integration with GitLab.

How it works

Describe — Type what AI product you want (e.g. "AI Legal Assistant for Indonesian law firms")
Build — Multi-agent engine provisions UI, backend, Cloud SQL, Storage, Auth automatically
Deploy — Instant public URL on *.autokeren.app
Integrate — Add MCP servers (GitLab, etc.) so your deployed agent can read repos, create issues, act autonomously
Monitor — Real-time SSE activity logs show every reasoning step transparently

Tech stack

Google Cloud Run — Serverless hosting & scaling
Google Cloud SQL (PostgreSQL) — Managed database per project
Google Cloud Storage — Asset & file storage
Gemini 3.5 / 2.5 Pro — Multimodal AI reasoning
GitLab MCP Server — @modelcontextprotocol/server-gitlab for autonomous repo actions
AutoKeren Studio — Visual application builder & deployment pipeline