Attuna
- Product
- Creative context for AI coding agents
- Role
- Founder & Product Designer
- Engagement
- Own productDesign & build
- Timeframe
- 2026
Giving AI coding agents a designer’s taste.
Attuna is a reference library and taste profile for designers. You collect the references you love, say what you like about each one, and connect AI coding agents such as Claude Code and Codex over MCP, so they can follow your creative direction.
Coding agents can build software, but their default visual output lacks the judgement and point of view of an experienced designer. Reference tools store inspiration and agents generate software, but nothing connected a designer’s own taste to the agents they already use.
My role
Attuna is my own product. I’m designing and building it end to end: the product and interface design, the Next.js app and the MCP server that agents connect to, and the waitlist site at attuna.design (opens in a new tab).
Key decisions
Let the designer’s own agent do the thinking
Attuna stores and structures each designer’s context, and their connected agent interprets it using the model access they already have. That keeps the product focused on taste rather than AI credits, and means free users don’t carry an inference cost.
Send real images, not descriptions
Over MCP, agents retrieve the actual saved images alongside the designer’s notes, design attributes and principles. That way they reason about the references themselves, not a summary of them.
Where it stands
The waitlist is open at attuna.design (opens in a new tab). The core MCP connection works end to end: an agent can retrieve a designer’s taste profile and real reference images, and use them to build UI.
Selected screens







