Roux

Inspiration

Great AI plans start with great context—but the people who hold that context are often not in the room when the plan is created. Teams know the history, risks, constraints, and open questions behind a project, but that knowledge is usually scattered across conversations and documents.

We built Roux to make that knowledge easier to share with Codex.

What Roux does

Roux interviews a team through a simple conversational experience. Each person can share what they know, what concerns them, and what decisions still need to be made. Roux then brings those insights back into Codex, where they can shape clearer, more complete plans.

The result is a planning workflow that combines AI speed with the knowledge of the people doing the work.

How we built it

We used Codex and GPT-5.6 throughout the build to explore the product idea, shape the user experience, write and refine the interface, develop the interview flow, and create the synthesis experience.

We focused on making Roux feel simple and approachable: one link for the team, thoughtful questions, clear progress, and a useful summary at the end. We also created a visual identity and a short product video to explain the experience quickly.

What we learned

The most important part of an AI workflow is often not the final output - it is the context gathered before the work begins. A few well-designed questions can reveal risks, assumptions, and decisions that would otherwise remain invisible.

We also learned that collaboration does not need to feel like a meeting. Roux lets people contribute asynchronously and turns many individual perspectives into one shared starting point.

Challenges

The main challenge was balancing depth with simplicity. Roux needed to ask meaningful questions without feeling like a long survey. We iterated on the wording, flow, and presentation until the experience felt conversational while still producing structured, useful insight.

We also had to design the handoff between the team interview and Codex so that the information felt actionable rather than just another transcript.

What’s next

We want Roux to support more project types, richer synthesis, and deeper connections to the tools teams already use. Our goal is to make shared context a natural part of every important AI-assisted project.

Built With

• 5.6
• 5.6sol
• chatgptwork
• codex.