BARCO Intelligence is live — one message, 23 tools, real Kubernetes After weeks of building on top of a live trading infrastructure, we just submitted BARCO Intelligence to All Things Agentic.
What it does One natural language message triggers a full autonomous workflow:
Queries 23 live production tools simultaneously (asyncio.gather) Writes and runs real Python on a Kubernetes cluster via Aider Publishes validated HTML artifacts to public MinIO URLs Self-validates every upload — minio_read_object fires automatically after every minio_upload_webpage, zero user input Commits the result to cross-session persistent memory (Ancora) The stack Gemini (gemini-3.5-flash) via Google GenAI SDK — native function calling, 12-round agentic loop Google Cloud Run — serverless orchestrator, scales to zero, built-in audit trail via Cloud Logging Gemma 3 on-device via flutter_gemma — offline fallback, zero downtime when Cloud Run is unreachable Real execution — recorded today
{ "tools_used": [ "get_memory_health", "get_swarm_health", "get_bridge_status", "minio_upload_webpage", "minio_read_object", "ancora_add_observation" ], "failed_tools": [], "model": "gemini-3.5-flash" } minio_read_object fired automatically — that's the QA self-validation loop at work.
Try it now No API key. No setup.
GET https://barco-gemini-agent-96738061556.us-central1.run.app/health POST https://barco-gemini-agent-96738061556.us-central1.run.app/chat {"message": "Audit this infrastructure and publish a live report.", "history": []} Wait 60 seconds — the response contains a live public URL with real infrastructure data.
Log in or sign up for Devpost to join the conversation.