Panasonic Well

Strategic foundation work the company recognized as its own.

THE BRIEF

In April 2022 I joined Yoky Matsuoka — formerly CTO at Nest, a co-founder at Google X — to scale Yohana, a concierge service helping families offload day-to-day tasks. It was running in a Seattle test market, but the real fact about it was that it sat inside a Panasonic incubator. Expand the market, roll out nationally, launch in Tokyo.

That brief held for three months.

THE REFRAME

The product couldn’t scale to what the business needed. So in July 2022 we mapped the consumer technology landscape on two axes: Mediated-to-Unmediated, Generic-to-Tailored. The crowded middle held the assistants and trackers — Google Home, Fitbit, Siri, Alexa. Yohana plotted upper-right with Uber. But that corner was crowded too, and competing there meant fighting well-capitalized incumbents for incremental advantage.

The move was a third axis the framework didn’t show. Every product on that grid was built for the individual user. Apple Home optimized one person’s preferences. Fitbit measured one person’s health. The entire category had organized around personalization for one. The whitespace was Collective — building for the family as a unit rather than the individuals inside it.

That became the foundation: the first consumer technology product organized around Collective Well-Being, on the premise that any family member’s wellbeing is structurally entangled with the others’.

THE RECOGNITION

In September 2022 the team traveled to Osaka and Tokyo for the Japan launch. The trip included the Panasonic Museum.

Kōnosuke Matsushita founded Panasonic in 1918 on the premise that business exists to serve the collective wellbeing of the communities it serves. He wrote a 250-year plan dividing the future into ten 25-year epochs, each one an opportunity to reimagine the business against the technology available at the time. Reading his original writings in translation, I recognized that the foundation we had articulated two months earlier was a rearticulation of what he had written 104 years before.

We hadn’t been told about his philosophy. We’d arrived from consumer insight and category analysis. That it matched is why Panasonic could absorb the work into its institutional DNA — the argument didn’t have to be translated into something the company could accept, because it had been Panasonic from the beginning.

WHAT GOT BUILT

By October 2022 we were designing an agentic platform rather than a concierge service with software around it: an architecture where AI agents acting on behalf of a family could work across health, time, education, and decisions, against a shared family profile rather than a set of individual ones. ChatGPT launched publicly six weeks later. The work wasn’t a response to the AI moment; it was already inside it.

The 2023 consumer insight program produced the operating test: families want Harmony, and every agent action had to Build Trust and Support Us. That gave the team a way to evaluate whether what they were building was the right thing. A full prototype landed in March 2024. Umi — the first agentic AI platform for families — was unveiled at the CES 2025 opening keynote, alongside the Panasonic–Anthropic partnership I co-architected. The work generated a family of patent applications on bringing agentic AI to families; I’m a named inventor on one, covering proactive AI systems.

THE INSTITUTIONAL OUTCOME

Umi launched into Panasonic Go — the company’s strategic reset, and the articulation of its fifth 25-year epoch. Go is the phonetic five in Japanese. I drove that strategy: Collective Well-Being was what connected the fifth epoch back to the first.

The operating entity wound down. The strategic intent is carried forward by the office of the Chief Strategy Officer.

WHAT IT PROVED

The most consequential strategic work doesn’t introduce a new argument to an institution. It produces one the institution recognizes as its own. The same method that found the Collective axis here found, at Avenue M, that a speed problem was actually a defensibility problem: start from the end state, and look for the axis the existing frame doesn’t show.