Four trainings in, the retailer booked three more and set up its own trainers.

Starting point

The retailer's product areas use Java, TypeScript on AWS Lambda and SAP, and maintain a legacy codebase of several million lines. In September 2025 we started with a discovery.

What we did

Benedikt Stemmildt led the work and consulted one day a week. Trainers from our network delivered the team trainings and the bootcamp.

• September 2025: a discovery. From October 2025, a weekly expert group on developer efficiency. In December 2025, agentic engineering workshops for the integration platform.
• January to June 2026: one consulting day a week, with an enablement kickoff in February.
• May 2026: four team trainings, for two Java groups, one TypeScript group on AWS Lambda and one product group.
• July 2026: a day-and-a-half remote bootcamp on a context engine, which produced a Graph-RAG prototype over several million lines of legacy code.
• August and September 2026: we trained the retailer's own trainers, so they can take the training to more teams.
• September and October 2026: two-day trainings for two developer groups in e-commerce and a remote training for the SAP team.

What changed

• In October 2025, developer efficiency was the topic of one weekly expert group. By October 2026, engineering, product and SAP teams had been trained in agentic engineering, and the retailer had its own trainers to carry on.
• After the first four trainings, the retailer booked more: two developer groups in e-commerce and the SAP team.

Based on our own engagement records, September 2025 to October 2026.

What this case does not show

No figures, because the client has not confirmed a measured result for publication. No quote, because the client has not approved one.

What we proposed

In September 2025 we proposed a federated context strategy: context engineering planned across the whole organization.