Danske Bank
Danske Bank rolls out 12 internal generative AI tools to its staff
Danske Bank has launched 12 internal generative AI solutions, led by the DanskeGPT assistant, and says almost 16,000 employees use at least one of them each month. The bank also reports that 90% of its employees have taken generative AI training.
Key figures
Copied exactly from the sources. Copy any figure with its citation.
Almost 16,000
Employees using one or more internal generative AI solutions each month, as reported by the bank in June 20251
90%
Share of all Danske Bank employees who have taken part in generative AI training1
12
Generative AI solutions launched by June 20251
What it took
Cost, team size and timeline were not disclosed in the sources.
The problem
Danske Bank, a large Nordic bank, wanted its staff to work faster on everyday tasks such as writing, summarising and answering internal questions. The bank also wanted a tool adapted to its security requirements, so that no data flows back to the model provider, and it needed people to actually use whatever it built.
What they built
The centrepiece is DanskeGPT, an internal assistant that helps staff write emails, prepare reports, translate, brainstorm and code. The bank has since launched 12 generative AI solutions in total. Besides DanskeGPT, it names DanskeAssist, GovDoc Assistant and an HR Assistant. New ones are tested and launched on a rolling basis.
How they did it
According to the bank's March 2024 release, DanskeGPT gives access to a version of OpenAI's ChatGPT 4.0 that is adapted to the bank's security requirements, so that no data flows back to OpenAI. The same release lists Microsoft 365 Copilot and GitHub Copilot as proofs of concept at that time. The bank ties the work to its Forward '28 strategy, which also covers a move to public cloud and a new data platform. By June 2025 it had named a Chief AI Officer and Head of GenAI to accelerate its efforts to integrate generative AI into its ways of working and improve the customer experience; the release describes the bank's focus since DanskeGPT launched as adoption, upskilling and training.
Results
In June 2025 the bank said almost 16,000 employees use one or more of its internal generative AI solutions every month. It also said 90% of all employees have taken part in generative AI training. The bank gives no savings or productivity figures, so the evidence is about adoption, not business impact.
Lessons for your project
Pair the tool with training. Danske reports that 90% of employees took part in generative AI training, alongside the launches.
Launch a general assistant first, then add narrower tools. Danske began with DanskeGPT and later added tools such as an HR assistant.
Put one senior owner on adoption. Danske named a Chief AI Officer and Head of GenAI, with a stated aim of accelerating how generative AI is built into its ways of working.
Track monthly active users, not only launches, so you can see whether staff actually use what you built.
Caveats
- Both sources are Danske Bank press releases. The figures are the bank's own and no outside party confirms them.
- The bank reports usage and training numbers only. It gives no figures for time saved, cost or quality.
- Only DanskeGPT has a named model: the March 2024 release describes it as a version of OpenAI's ChatGPT 4.0, adapted to the bank's security needs. The model behind the other solutions is not stated and may have changed since 2024.
- The releases give no exact launch date for DanskeGPT. The deployed date shown is the month of the March 2024 release that says it was already released to employees.
- The June 2025 release says only that DanskeGPT launched a little over a year earlier. The monthly user count covers all 12 solutions together, not one tool.
Sources
- Danske Bank further strengthens Generative AI focus with new leadership appointment
Danske Bank · 2025-06-16 · Press release · danskebank.com
- Danske Bank Increases Investments in Generative AI
Danske Bank · 2024-03-14 · Press release · danskebank.com
Summarized in our own words from the sources above. Spotted an error? See corrections.