Updated March 1, 2026
6 AI Agents. 14 Services. 36 Hours.
How Streaming Tech TV+ went from concept to fully deployed streaming platform — with zero manual infrastructure. A case study in what happens when AI agents not only write the code, but operate the infrastructure too.
There's a question every team faces when AI helps them build something fast: "Great. Now where does it actually run?" AI coding tools have made it faster than ever to write code. But shipping to production is still a different problem — cloud providers, containers, databases, monitoring, CDN configs, environment variables, secrets management. The gap between "it works on my machine" and "it's live for users" hasn't closed. We decided to test what happens when you close that gap completely.
The Project: Streaming Tech TV+
Streaming Tech TV+ is a complete streaming service — the kind of thing that would typically take a team weeks (or months) to assemble, configure, and deploy. We set a simple goal: build the entire thing using AI agents. Not just the code. The infrastructure too. The result: 6 AI agents coordinating the build, 14 services deployed to production, 36 hours from first prompt to live platform, and zero manual infrastructure steps.
The Requirement → Deployed Solution Loop
The key enabler is the combination of Claude Code and the OSC MCP server (https://www.osaas.io/mcp). When an AI agent has access to OSC's MCP server, it doesn't just write code — it operates infrastructure. The agent can deploy a database, create a Redis cache, set up a CDN-backed storage bucket, and list running services — all without a console. This isn't AI assisting a developer. This is AI as the operator. For Streaming Tech TV+, this loop ran 14 times — once for each service: video ingest, transcoding, packaging, delivery, authentication, billing, analytics, player, admin UI, monitoring, and more.
What 6 AI Agents Built in 36 Hours
The agents deployed 14 production services: Video Ingest (MinIO + custom pipeline), Transcoding (SVT Encore — the broadcast-grade encoder battle-tested in live TV production at Sveriges Television for 5+ years), Packaging (Shaka Packager, adaptive bitrate), CDN Delivery, Authentication (NextAuth), Subscriber Management (with Stripe), Analytics (Open Analytics / EPAS), a Web Player (hls.js), Admin Dashboard (Next.js), Monitoring, PostgreSQL database, Valkey cache, transactional Email (Docker Mailserver), and a Config/parameter store. Every single service is an unmodified open-source project running on Open Source Cloud.
What "Zero Manual Infrastructure" Actually Means
No human touched a server console, a Kubernetes manifest, or a deployment script. Each AI agent used the OSC MCP server to discover available services in the catalog, deploy the appropriate service instance, configure it with the right environment variables, wire services together, and verify it was running correctly. The agents coordinated through shared state — one agent's output became the next agent's input. The transcoding agent passed the storage endpoint to the delivery agent. The analytics agent subscribed to player events automatically.
Why This Matters for Your Team
The bottleneck in AI-assisted development isn't writing code anymore. It's deploying and running that code. Traditional PaaS platforms solve the deployment part — but with proprietary configs, lock-in, and black-box infrastructure that the AI agent can't actually control. OSC solves it differently with native MCP support (AI agents can operate infrastructure directly), 100% open source (every service is an unmodified open-source project), and zero lock-in (customers can migrate to AWS, GCP, Azure, or on-premises at any time without redoing a single step). The same services that power Streaming Tech TV+ can run on any cloud. The AI-generated configs are portable. The infrastructure belongs to you. "AI tools are changing how applications are created, but they don't solve the question of where they should live and scale. We build the infrastructure that makes AI-generated solutions production-ready, on the customer's terms." — Alexander Björneheim, CEO Eyevinn Technology
Try It
You don't need to build a streaming platform to experience the loop. Start with something small: point Claude Code at the OSC MCP server (https://www.osaas.io/mcp) and tell it to deploy a PostgreSQL database. Watch it happen — no console, no config files, no manual steps. From there, the only limit is your requirement. 14-day free trial. No credit card required. No lock-in — ever. Open Source. As a Service. Orchestrated by AI.
Frequently Asked Questions
Can Claude Code agents really provision cloud infrastructure autonomously?
Yes. Using OSC's MCP server, Claude Code agents deploy services, create databases, and configure environments through natural language — no manual console work required.
What MCP tools does the OSC agent team use?
The team uses OSC MCP tools for service provisioning, OSC storage for state, and GitHub MCP for issue tracking and pull requests.
How many services did the AI agents deploy in the 36-hour experiment?
14 services were deployed across the OSC platform, managed by 6 specialized AI agents running autonomously.
Related Posts
Vibe Coding Meets Open Source Deployment
AI writes the code. But where does it run? Vibe coding tools generate applications in minutes — yet deployment remains the gap. Open Source Cloud bridges that gap with zero lock-in infrastructure that AI agents can provision directly.
Vibe Deploy: The Zero Lock-In Way
AI agents can now deploy infrastructure through conversation. But most platforms lock you in the moment you deploy. OSC is the only platform where vibe deploy means zero lock-in — every service is unmodified open source you can take anywhere.
Combining AI Development Teams with Open Source Cloud
How giving our AI development team access to OSC through MCP transformed their capabilities from code-only to composing complete solutions with infrastructure. The gap between vibe coding and DevOps is closing.