YASNO (DTEK)
YASNO cuts customer inquiry handling time from 4.5 to 3.5 minutes with Azure OpenAI
YASNO, the energy retail brand of Ukraine's DTEK group, gave its contact centre agents an AI assistant that suggests answers to customer inquiries. Microsoft's case study reports that average handling time fell from 4.5 minutes to 3.5 minutes.
Key figures
Copied exactly from the sources. Copy any figure with its citation.
22%
Reduction in inquiry handling time, as stated in the headline of Microsoft's story1
more than 300
Customer inquiries the contact centre handles daily1
What it took
Cost, team size and timeline were not disclosed in the sources.
The problem
YASNO is the brand under which DTEK provides retail electricity and gas supply in Ukraine, serving about 3.5 million individuals and businesses. Its contact centre handles more than 300 customer inquiries every day. The story says agents were taking a lot of time processing each request, which slowed operations and hurt efficiency and customer satisfaction.
What they built
YASNO built an AI assistant that gives agents real-time recommendations while they handle an inquiry. It sits in a visual panel inside the Cisco contact centre software agents already use. The assistant is built on Azure OpenAI Service, Microsoft's hosted version of OpenAI's language models, and Azure AI Search, which finds the relevant records.
How they did it
The assistant pulls information from billing and accounting systems, knowledge bases and other master data. The work was supported by DTEK's IT division, MODUS X. The story describes a successful pilot followed by an official launch in the contact centre, without dates. It adds that DTEK had already used Microsoft Power Platform for debt management.
Results
According to Microsoft's story, average handling time for an inquiry dropped from 4.5 minutes to 3.5 minutes. The headline calls this a 22% reduction.
The story also says the assistant is expected to process 80% of inquiries automatically. That is an expectation, not a measured result. Cost, team size and timeline are not given.
What to copy
Put AI suggestions inside the screen agents already use, connect them to your own customer records, and pilot with a small group before launch. Measure handling time before and after, and record how you measured it.
Lessons for your project
Keep the assistant inside the tool agents already use. YASNO built a visual interface inside its existing Cisco contact centre software, so agents did not switch screens.
Run a pilot before the official launch. YASNO ran a pilot first and then launched the assistant officially in the contact centre.
Connect the assistant to your own records. It draws on billing and accounting systems and knowledge bases, so its suggestions fit each customer's case.
Build on earlier groundwork. DTEK already used Microsoft Power Platform for debt management, which the story says made adoption smoother.
Caveats
- The only source is Microsoft's customer story. The figures are reported by the vendor with YASNO and are not independently verified, and the story does not say how handling time was measured or over what period.
- The page shows the date 4/01/2025, which can be read as 4 January or 1 April. Only the year 2025 is recorded (required field), taken from the page date and not from a go-live date, which the story does not give.
- The story does not name the OpenAI model. The model is recorded as undisclosed, and Azure OpenAI Service is listed as the platform.
- The story says the assistant is expected to handle 80% of inquiries automatically and mentions a possible customer-facing assistant. These are expectations, not results, and are not counted here.
- The 22% figure appears only in the headline. The body gives the 4.5 and 3.5 minute figures.
Sources
- DTEK's YASNO boosts customer service with Microsoft Azure OpenAI, reducing inquiry handling time by 22%
Microsoft · Undated · Vendor case study · microsoft.com
Summarized in our own words from the sources above. Spotted an error? See corrections.