A fast-growing DTC wellness brand

The challenge

A fast-growing DTC wellness brand came to us with a familiar question: how do we get more out of AI? They had roughly 15 people, a handful of tools already in play (Claude, some automations, a CRM integration) and a Head of Ops who was very good at her job. The tools were being used reactively and in scattered ways. Nobody had stepped back to ask where AI would actually make the biggest difference.

The answer was sitting inside a task so unglamorous it had barely registered as a problem.

Every week, the ops lead spent hours chasing around 30 suppliers by email: checking ETAs, confirming deliveries, keeping production partners informed. When it worked, nobody noticed. When it slipped, the consequences were serious. A missed update from one ingredient supplier could mean a delayed delivery to the manufacturer. The manufacturer’s next slot might be two weeks out. That is a stockout: lost revenue, unhappy customers, and damaged relationships with production partners who valued reliability above everything else.

The other problem was structural. Supplier status lived in her inbox and her head. She was the only person holding the full picture. If she was away, nobody could pick it up.

The approach

We ran a half-day operational diagnostic on site, followed by weekly advisory.

We did not start by asking what AI could automate. We started by asking what mattered most to the business, and what went wrong when it went wrong.

In the morning, we mapped core operational processes end to end: every step, every handoff, every workaround. Then we traced the consequences of failure backwards. Stockouts were the catastrophic outcome, and the chain causing them ran through supplier management, production scheduling, financial approvals, all the way back to something as mundane as expense receipts not being submitted on time. This surfaced three leverage points. Supplier management came first, not because it was the most obviously automatable, but because it combined a significant weekly time cost with the worst downstream risk.

In the afternoon, rather than building the solution for her, we coached her to build it herself. We talked through the building blocks (how to set up projects, scheduled automations, email connectors) and she started building while we were still in the room. The architecture was straightforward once the right problem was identified: a scheduled agent that checks for supplier updates, drafts chaser emails, and produces status briefings.

In weeks one and two, she tested the agent against real supplier communications, refined how it handled different scenarios, and built genuine confidence in what it could and could not do. Weekly advisory meant blockers cleared fast.

By week three, with the agent working reliably, she documented it, wrote guides for the team, and ran training with every department head. They started building their own automations.

The results

The most immediate gain was more than two hours a week saved on supplier management, with the biggest saving on Mondays, previously the most operationally intense day. Proactive chasing meant fewer missed communications and less risk of production delays. A shared tracker replaced the “it’s all in my head” problem: the whole team could see supplier status for the first time.

But the more important result was the multiplier. In roughly four weeks, the ops lead went from basic AI user to building structured automations, writing team documentation, and training every department head. Department heads began running their own AI sessions. The business went from scattered, reactive AI use to something strategically directed.

What made it work

The secret was in knowing where to start. Most businesses adopting AI start by looking at what is repetitive, or what AI is obviously good at. We started from the other end: what has the biggest consequences when it goes wrong? Starting from consequences, not capabilities, is what got us to the right problem quickly.

And the coaching model mattered just as much as the diagnosis. It would have been faster to build the automation for them. But building for someone creates a dependency. Coaching them to build it creates a capability.

The smartest investment is not in building automations. It is in making your best people exceptional, because they will make everyone else better too.