Qwen War Room

OPS REAPER MISSION CONTROL

Inspiration

Qwen War Room was inspired by the way real high-pressure teams make decisions: one person should not carry the whole load alone. Good operators use advisers, challengers, builders, and verifiers before committing to action.

The project turns that pattern into a Qwen Cloud-powered multi-agent command council. A user brings a messy business problem into the War Room, and a set of specialist agents break it down, challenge assumptions, identify risks, propose actions, and generate an executive brief.

What it does

Qwen War Room is a multi-agent decision cockpit for business operators.

The system takes an input brief and runs it through a council flow:

Intake captures the problem.
Strategist turns the problem into a structured plan.
Red Team challenges the plan and surfaces risks.
Builder proposes an execution path.
Verifier checks the plan against the original objective.
Chair synthesizes the final command brief.

The demo scenario uses “Hot Brew Crew,” a fictional small business, to show how the War Room handles real operational pressure without exposing private client data.

How we built it

The project is built around Qwen Cloud models, with Qwen3.7 Max intended for the Chair/synthesis role and Qwen3.7 Plus intended for worker agents such as Strategist, Red Team, Builder, and Verifier.

The application uses a lightweight web stack:

Node.js / Express backend
Simple frontend interface
Qwen Cloud API integration
Environment-based API key handling
Structured agent routing
Markdown-style report output

The goal was to keep the system fast, inspectable, and demo-ready rather than overbuild a platform.

Challenges we ran into

The biggest challenge was compression: turning a larger command-system idea into something that could be demonstrated clearly within the hackathon window.

We also had to separate what matters for the demo from what belongs in a later production version. Features like voice input, OBS integration, persistent memory, and multi-user rooms are valuable, but they were intentionally kept out of scope for this submission.

What we learned

We learned that agent systems become much more useful when each agent has a bounded job and a clear handoff. The value is not just “more AI agents,” but structured disagreement, verification, and final synthesis.

Qwen Cloud is a strong fit for this pattern because different model tiers can be assigned to different roles: stronger models for synthesis and cheaper models for worker tasks.