AWAI

01What AWAI is

After an hour-long conversation with an AI, almost no one can recall exactly what they were thinking the whole time. The ideas that surfaced midway, the threads picked up and the ones let go, the branches unconsciously cut to get back on track — because a conversation flows along a single line in time, the structure of “what the whole thing was really about” stays invisible, even in hindsight.

AWAI takes this content that would otherwise flow past and vanish, and reorganizes it by meaning. It automatically extracts each individual thought from the conversation and places related ones close together, distant ones far apart — rearranging everything not by chronological order but by “distance in meaning,” while also summarizing the key point of each. The whole picture of a conversation, invisible from simply reading it back, becomes something you can take in at a glance.

That whole picture is shown as a view you can explore in three dimensions. The 3D look catches the eye, but it is only a way of presenting things — a means to grasp it intuitively. The heart of the value lies in the fact that the conversation has been organized by meaning, down to its key points.

What works especially well is being able to find the threads that stalled off to the side. “Whatever happened with that?” — you rediscover a point you had cut loose and forgotten, and can now consciously decide what to do with it. This is AWAI’s most distinctive value.

02Five features — one core, and its applications

AWAI looks like a single service, but at its center is one core engine, from which features of different characters branch off. Normally, building just one of these would be a business in itself.

Thought-Process AnalysisCore

Just paste a conversation with an AI, and it organizes the content by meaning and visualizes it down to the key points. The core feature where forgotten points resurface — the prototype for everything else.

Group AnalysisPillar

Have every member of a team or organization talk to the AI on the same topic, and it organizes the distribution of everyone’s opinions into a single view (see 03).

AWAI LivePillar · in dev

Structures a meeting in real time into a “map of the discussion,” and lets people who speak different languages talk in their own native tongues — a new kind of meeting tool (see 04).

Thinking-Type & Compatibility DiagnosisEntry

Talk with the AI for a few minutes, and it expresses your thinking tendencies, or how two people’s ways of thinking fit together, as a character and a set of axes.

03Group Analysis — a “map of opinion” for an organization

One of the practical pillars, and the feature whose value is clearest in a business setting. Think of it as the next evolution of the survey. It resolves all three of the structural weaknesses a traditional survey carries.

It is not merely a tallying tool. It is a device that draws out “what the members think, and how” — with no steering, no manual tallying, and no subjectivity. It reveals the very distribution and overlap of opinion that the numbers in a survey could never show.

04AWAI Live — run a meeting while watching it unfold

※ This feature is currently in development, in preparation as AWAI’s next pillar.

Where an ordinary record transcribes words in the order they were said, AWAI Live groups the discussion by topic as people are still talking.

Combined with translation technology (Gemini Live Translate), people who speak different languages can discuss seamlessly, each in their own native tongue. A Japanese speaker and an English speaker talk without an interpreter in between, sharing the same map of the discussion — we are building the very place where, across the barrier of language, people can think together as they talk.

05Scale — how much substance there is

Setting aside extras and filler, AWAI’s substantive implementation runs to about 160,000 lines — the heft of a serious, long-lived piece of software. And it was built by one person, in about three months. A scale that would normally take a specialist team several years was assembled by concentrating on the design and the precision of interpretation.

This speed was made possible in part by recent advances in AI-assisted development. But what AI speeds up is only the “assembly work.” AWAI’s heart — the design of how to read a conversation and organize it by meaning, and the precision honed on real data — is the part that can only be built up over time.

06Quality — how carefully it is built

Being large does not mean the inside is not hollow. To check, we had an AI (Claude) inspect the code from a third-party perspective, and the result was 8.5 out of 10. For non-technical readers, just three points:

• Robust handling of money — built with great care so there are no double charges or missed transactions; you can entrust customers’ money and data to it with confidence.
• Thorough automated checking — a system that automatically inspects “whether anything is broken” is built in, in roughly the same volume as the product itself.
• Few shortcuts — there are very few stopgap “just make it work” patches, so it is built to be maintained over the long term.

The interface, too, fully supports both dark and light modes and stays consistent across every screen. And because it runs entirely in either Japanese or English — not just the menus, but the very content the AI returns — there is no need to rebuild it to take it abroad.

07Why it can’t simply be copied

Copying an “app that displays a conversation” is possible. But AWAI’s real value lies not in its appearance, but in the precision with which it reads a conversation and organizes it by meaning.

That precision was built by trying it again and again on real data, layering small adjustments. And it is precisely this accumulation of adjustments that you cannot catch up to later by compressing time. However capable a team, however serious the attempt, what they end up with stops at “something that looks the part” — and that is why it is not easily overtaken.