There’s a pattern emerging in software: take an existing product, add a chat interface, call it AI-powered.

We went the other way.

Built around agents, not bolted on

Gotrino isn’t a design system tool with AI features. It’s an agent-first platform where specialist agents are the foundation, not an afterthought.

What does that mean in practice? The dashboard, the CLI, future integrations—they’re all interfaces to the same underlying agent system. The intelligence isn’t in the UI. It’s in the agents.

Specialist agents, tightly scoped

We’re not building one general-purpose AI that tries to do everything. We’re building specialist agents, each focused on a specific domain: accessibility, performance, token health, documentation coverage.

They’re tightly scoped. They know their boundaries. And critically, they cite their sources—scan results, source files, documentation. No hallucinated explanations. No plausible-sounding guesses.

Intent, not commands

One thing AI is genuinely good at: understanding what you mean.

When you ask “how’s the Button component doing?”, you’re not running a command. You’re expressing intent. The system figures out which agents to involve, what data to gather, and how to synthesise a response.

This matters because design system questions are rarely simple queries. They’re contextual, nuanced, sometimes vague. Intent recognition lets us meet you where you are.

Beyond the chatbot

AI doesn’t have to mean a chat interface. We’re exploring how to surface agent knowledge in ways that fit naturally into workflows—health dashboards, PR comments, CI checks, proactive notifications.

The question isn’t “how do we add AI?” It’s “what problems can agents solve, and what’s the right interface for each?”

Why this matters

Agent-first architecture isn’t about being trendy. It’s about building something that can genuinely help—with the transparency and reliability that design system teams need.

We’re not interested in AI as a checkbox. We’re interested in AI that earns trust.