Nutrition Geeks is a fast-growing UK nutrition brand. It put an Adam AI agent team to work on its own private hardware, starting with customer service.
- 15x growth: from 10M to 150M in a little over 12 months
- 80% less workload in some departments
- Customer service workload down to 20% of its previous level once AI was switched on
The challenge
Nutrition Geeks was growing fast across several warehouse sites, with a large customer service team handling a rising volume of orders, amendments and cancellations. The business wanted the speed of AI without handing customer and order data to public AI tools.
The approach
- AI Readiness first: IT environment, security controls and access reviewed before any agent went live
- Private AI on site: two AI servers in the business running the full Adam AI stack, with its own private knowledge base
- Data stays on site: customer and order data never leaves the business
- Fully self-sufficient: the platform runs on the client’s own hardware, under their own control
The first agent: customer service
The first digital worker was a customer service order advisor, trained on the team’s own operating instructions. It runs the same verification checks as a trained advisor, offers an amendment before a cancellation, checks the warehouse system before acting and never cancels an order that has already shipped. It never issues refunds itself: they go to a human review queue, and it escalates to a person when a system does not respond.
People stay in control
Refunds and anything irreversible wait for a person to approve. Every action is logged.
Why it worked
The agents learn how the team works and keep that knowledge, like permanent staff rather than a temp starting from scratch every day. With the first department proven, digital workers are now being rolled out across the business.