Water Street Collective · FMCG for human performance

Fitness in one app, nutrition in another, recovery in a third, mindfulness somewhere else again. The person moving between them is the only thing holding the day together.

Water Street Collective wanted to build for the day rather than the category: a system which knows whether someone is at work, in training, or asleep, and offers the right thing accordingly.

The research ran through days rather than demos: how a person moves from desk to training to sleep, which products they reach for at each turn, and what they expect those products to already know about them. Nobody lives in categories. They live in days.

The research pointed somewhere unfashionable. People no longer read tailoring as a luxury, they read it as the baseline, and a product which cannot adapt now reads as broken rather than basic. Human performance is holistic, which means the data has to be too.

So we mapped the whole ecosystem with the person at the center. Channels sit by proximity rather than by category, from the sensors worn on the body out to the regulators at the edge. What sits closest is read most often. What sits furthest sets the constraints the rest has to respect.

That set the architecture before it set the interface. Integrate across touchpoints, close the loop, and let each interaction refine the next.

The defined experience is a loop a person can watch close: the day arrives as one reading, the reading becomes a suggestion, the suggestion is answered with a yes or a no, and the answer tunes tomorrow. The product learns the person, and the next products start from evidence.

We built the proof of concept for a small group rather than a market. Its job was to mature the algorithm against real days: a strategic onboarding experience, a homescreen built from the data, timed check-ins, detailed insights, and a first pass at recommendations. It looked plain on purpose. Design was kept deliberately minimal: the point was a functioning build which could learn from real days, and the algorithm evolved with each one. The MVP came after the algorithm had something to stand on. The proof of concept carried three tabs. The MVP carried four, and the new one was Goals.

Personalized journeys driven by an adaptive homescreen. Targeted nutrition, breathwork, and mindfulness in the same place. A sensorial layer designed to capture and display state of being, so the product can show a person their own day back to them.

Built on modular microservices with low-latency APIs, scaled for millions of IoT data points, GDPR compliant by construction. Every screen and every state was drawn once, so engineering built from one map rather than from a set of opinions. Personalization was validated with PANAS, the Positive and Negative Affect Schedule, rather than by assertion.

The homescreen carries the whole state, and the color behind it moves as the day moves. Goals track sleep, focus, and energy against a target. Recommendations turn the reading into something to do: fifteen minutes of yoga, a cup of tea, five minutes of breathing. Data on the way in, a decision on the way out.

There is no headline percentage on this page, because the agreed measures here are qualitative and we will not dress them up.

What the work delivered is a loop a person can watch close. The next day the app asks a plain question: we recommended some activities, did you take them? Yoga, fifteen minutes, 16:45. Tea, one cup, 22:00. Yes or no. Then it says how it thinks that changed the day, stated as a belief rather than a fact. The answer feeds the next recommendation, and the recommendations assemble a box of products which fits what the data has learned. Points earned by following the advice pay for the box.

That is the argument for building this inside consumer goods rather than beside it. Consumer data reaches product development without a research detour, so the next wave of products starts from evidence rather than from a brief. The loop is the asset.

Record. Water Street Collective, London.

Work led by Formist’s founders in prior practice.

Discipline. Fast-moving consumer goods · connected products · applied AI.

Ownership. Studio-built, client-operated.

The MVP surface: the day as one reading, the goals beneath it, the lull the algorithm catches, and the recommendation which answers it.