About

intentfirst.ai

About intentfirst.ai

An open design language
for how AI should understand people.

AI is getting smarter at processing language. It hasn't gotten much better at understanding the person behind the words — where they are, what they're carrying, who else is in the room, what's actually at stake.

Context Grammar is an open framework for closing that gap. Token vocabulary, brain architecture, rule engine — all documented here, free to use, free to build on.

Takao Umehara

Takao Umehara

Founder, intentfirst.ai · Product designer

Over a decade building digital products at Fortune 100 scale. Most recently designing AI agents and experiences at a large telecom — from inside the AI era, noticing what design still hadn't solved.

I've spent years designing products used by millions. Most recently — agents and AI experiences at a large telecom — right as the AI era was arriving.

From inside that moment, I kept noticing the same thing. AI is still hard to use. In many places. In small ways and big ones. The smarter the model gets, the more visible the gap becomes.

AI keeps getting smarter at processing our words. It hasn't gotten any better at understanding our world.

That bothered me — not as a complaint, as a question I couldn't put down. What can a designer contribute toward a future that's almost here? I can't solve everything. But one piece — that, I can try.

So I started. From this moment — May 2026 — toward the slightly-near future. I'm a designer; designing the future is the work.

Context Grammar is what came out. Eight tokens that give AI a structured way to understand not just what someone said, but who they are when they said it. Some — physical state, social exposure, form factor — phones and smart devices can sense today. Some — cognitive load, what's really at stake — they can't, yet. I want to be honest about that.

But the future these tokens point toward — AI that fits the way people actually live, where humans can trust AI and AI can earn that trust — I believe in it. And I want to contribute one piece toward it.

This is my best attempt, so far.

Vision

A shared vocabulary for human context.

Everything here is open — the framework, the token specs, the research, the projects. The goal is a shared foundation that anyone can build on. Not a product, not a consultancy. A commons.

Get in touch.

Context Grammar resonates with your product challenges? I'd like to hear from you.