Shiseido Japan's 130 AI Ambassadors Took Gemini Adoption Past 80% — Now Comes the Hard Part

How Shiseido Japan Built Its 130-Person AI Ambassador Network

Shiseido Japan has spent about two years turning generative AI from an experiment into a routine part of daily work. The cosmetics unit, part of the group founded in 1872, now runs a network of around 130 “AI ambassadors” across about 40 departments. Ambassadors learn new tools, help colleagues use them, and feed practical lessons back to a central transformation team led by the DX and AIX Strategy Department. Executives outlined the approach at Google Cloud Next Tokyo in July.

The push is a direct response to three pressures named by the company: stagnant domestic sales, intensifying competition and a shrinking labour force. The rollout itself was deliberately slow. Management workshops began in spring 2025, tools were tested against real business tasks, and from early 2026 transformation managers held individual discussions with every department head. Each head was asked to nominate the person they trusted most to drive AI use in their team, producing the roughly 130 ambassadors.

Shiseido Japan calls the result a “sandwich” structure: direction from leadership on top, ambassador-led activity from the shop floor underneath. The company reports that more than 100 hands-on activities tailored to individual departments have run since April 2026, that Google Gemini's active rate among employees exceeds 80%, and that 40% of staff use the tools for work at least once every three days.

The current structure follows a bigger organisational shift. Shiseido Interactive Beauty (SIB), the digital strategy subsidiary created in 2021 as a joint venture with Accenture, ended its partnership contract in late 2025, became a wholly owned subsidiary in January 2026 and was absorbed into Shiseido on 1 June 2026. Shiseido says SIB delivered value in recruiting and developing digital talent and accumulating know-how; folding it into the main business is meant to speed decision-making and embed digital capability where daily operations actually happen.

Why Shiseido Japan Moved SIB's Digital Team Inside the Business

The logic behind folding SIB back into the parent

SIB's original purpose was to push change into Shiseido from outside, with Accenture's expertise and a separate mandate. Shiseido says the experiment produced talent and know-how, but it also left a gap between the people building digital capabilities and the business units meant to use them. Moving those functions inside Shiseido and Shiseido Japan shortens that loop. The sequence is familiar in large Japanese companies: prove a capability in a separate vehicle, then internalise it once it is credible enough to operate in the core business.

Why ambassador selection mattered more than tool installation

Transformation manager Hikari Kikuchi's team spent about a year on preparation, and the key step was asking department heads — not HR or IT — to choose the ambassadors. That transfers authority for change to the managers who own local performance. A top-down mandate alone tends to produce logins without behaviour change; a purely bottom-up enthusiast group can drift without direction. The sandwich model tries to create both channels: management sets direction, ambassadors create momentum in their own teams.

The central-team bottleneck is the real test

The headline numbers — an 80%-plus active rate for Gemini and 40% of employees using it for work at least once every three days — measure broad adoption of everyday assistance. But the company acknowledges that more advanced data-analysis requests are still flowing to a central specialist team, and that handling them individually has become a bottleneck. That suggests the limits of adoption metrics: they show how many people use AI, not how much specialised work has moved out of one queue. The reported next phase — teaching employees to solve data problems themselves rather than supplying answers — is the harder part of the transformation, and it is still early.

What Enterprise AI Teams Can Learn From Shiseido's Rollout

Shiseido's rollout is a useful case for enterprise AI leaders, mainly because its weak points are as instructive as its metrics:

• Sequence preparation before tools: management workshops ran in spring 2025, followed by testing on real work, before broad deployment began.
• Give department heads ownership: all around 40 department heads were asked to personally nominate the ambassador they trusted most, which aligned AI adoption with local credibility.
• Track behaviour, not just access: useful baselines are Shiseido's reported figures of more than 80% Gemini active rate and 40% of employees using AI for work at least once every three days.
• Plan for the central-team queue: with analysis requests concentrating in one specialist department, design the shift toward coaching and self-service before the bottleneck grows.