BKW
BKW builds Edison, a staff chatbot on Azure OpenAI, and speeds up media inquiries by 50%
Swiss energy and infrastructure group BKW built Edison, a generative AI chatbot for all of its roughly 12,000 employees, in eight weeks. Microsoft reports that media inquiries were processed 50% faster and that 8% of staff were active users two months after rollout.
Key figures
Copied exactly from the sources. Copy any figure with its citation.
The problem
BKW, a Swiss energy and infrastructure group with roughly 12,000 employees, wanted staff to use AI in their daily work in a way that complied with company rules. It did not want a tool for only a handful of people.
What they built
Edison, a chatbot for all employees. It combines generative AI with search over the company's own information, so staff can analyse internal documents in context and generate content. Its first use case was corporate communication, to speed up editorial work and answers to inquiries.
How they did it
BKW chose to build Edison with its own team, with Microsoft support, on Azure OpenAI Service and Azure AI Foundry. It took eight weeks. The team first tested specialised applications in small groups with a clear theme, planned a phased rollout, and listed over 40 further use cases in a documented pipeline. It is now testing agentic AI, meaning systems that carry out multi-step tasks on their own.
Results
Microsoft reports that media inquiries were processed 50% faster and that 8% of staff were active users within two months of the company-wide rollout. The story gives no baseline for either number.
Lessons for your project
Start with a narrow first use case. BKW began with corporate communication, then widened to all staff.
Test in small groups with a clear theme before opening to everyone.
Plan a phased rollout from the start, and track active use. BKW measured the share of staff using Edison two months in.
Build a use case pipeline early. BKW drew up more than 40 documented use cases.
Caveats
- The only source is a Microsoft customer story. The baseline and method for the 50% figure are not stated.
- The model is not named. The story says only that Edison uses Azure OpenAI Service and Azure AI Foundry. Azure OpenAI Service serves OpenAI models, so the vendor is recorded as OpenAI and the model as undisclosed.
- The story gives no launch date. deployedDate is the year of the story and a BKW staff magazine (issue 1/2025) already describes Edison as available to all staff, so the real launch may be earlier.
- No independent source was found in English or German.
Sources
- Edison on Microsoft Azure gives BKW a platform for hands-on AI
Microsoft · 2025-06-20 · Vendor case study · microsoft.com
Summarized in our own words from the sources above. Spotted an error? See corrections.