The "Impossible" Sprint:
5 Media Sites in 14 Days
The brief was simple: Build a media network from scratch in two weeks. No massive team. No QA bottlenecks. Here is the architecture that made it possible.
The Constraints
If you try to do this with a traditional CMS and human-heavy workflow, you will fail. The only way to hit this velocity is to stop thinking about publishing and start thinking about orchestration.
1. The "Control Tower" Architecture
We didn't build 5 websites. We built one Control Tower. Instead of a monolithic server, we utilized a "Supervised Actor Model." Every brief, draft, and email is an isolated process that never sleeps.
2. The Content Assembly Line
We don't brainstorm topics manually. We ingest Signals. When a signal hits a threshold, it triggers an autonomous workflow. Humans only touch the "Polishing" phase.
3. The Multi-Tenant Shell
We built one rigorous Next.js frontend (The Shell) and deployed it 5 times with different configuration files. This reduced frontend dev time by 80%.
The "Shell"
Shared Components, Layouts, API Routes, Analytics
The 14-Day Execution Plan
Week 1: The Engine (Backend)
Build ingestion engine to pull data streams. Setup Postgres/Schemas.
Connect Brief, Draft, and Fact-Check agents. Ensure fault tolerance.
Build audience graph and frequency capping logic (Backend only).
Week 2: The Surface (Frontend)
Build Design Tokens, Cards, PageHeaders. Ensure responsiveness.
Turn on the Control Tower. Agents generate 10 assets. Human review.
Map domains. Inject assets. Verify compliance gates.
Why This Works
We hit this deadline because we aren't relying on human memory. We are relying on a Supervisor System that never sleeps.
- If an API fails, the system retries instantly.
- If a draft lacks sources, the system rejects it automatically.
- If an email violates a rule, the system blocks the send.
We build the governance first, and the content flows through it. That is how you scale a network in a fortnight.