← Back to Nordea scales AI to more than 10,000 users on a shared model-agnostic platform built on AWS Bedrock

Source proof for Nordea scales AI to more than 10,000 users on a shared model-agnostic platform built on AWS Bedrock

Source-bound proof

Verified source excerpts for every supported field

Each colour maps a published value to the exact source passage used to support it. Only bounded excerpts are public; administrators can inspect the complete captured source.

10 fields supported

Title

Nordea scales AI to more than 10,000 users on a shared model-agnostic platform built on AWS Bedrock

quote · high
…ea, the focus is now on transforming workflows rather than building prototypes. Scaling to more than 10,000 users required discipline and patience. The work moved from laptop pilots to a production‑grade platform. It was delive…
…r Work (AIAW) podcast. Instead of scaling individual AI use cases in isolation, Nordea chose to build a shared, flexible platform leveraging Amazon Web Services (AWS) Bedrock that works with different models. This is a managed service from Amazon that provides access to a variety of larg…

Description

Nordea moved from small-scale chatbot prototypes to an enterprise AI capability, building a shared, flexible platform on Amazon Bedrock that works with different models to avoid vendor lock-in. Governance, including monitoring, access controls and compliance, is built into every layer. The source says scaling to more than 10,000 users moved the work from laptop pilots to a production-grade platform, delivered in stages.

quote · high
…as AI models evolve. Mattias Fras, Head of AI Adoption. "Not a magic solution" Governance is incorporated into every layer. This includes monitoring, access controls, and compliance. This approach makes governance a technical enabler rather than a barrier. As Ma…

Company

Nordea

quote · high
…era of chatbots feels ancient,” says Olof in the AI After Work (AIAW) podcast. Instead of scaling individual AI use cases in isolation, Nordea chose to build a shared, flexible platform leveraging Amazon Web Services (AWS) Bedrock that works with different models.…

Industry

Financial Services

classification · high
…ductivity and those that truly transform workflows. Nordea’s journey shows that responsible AI in banking is built on strong building blocks, built‑in governance and organisational change, even in a regulated environment…

Problem

Nordea moved from small-scale chatbot prototypes towards an enterprise AI capability, choosing to build a shared platform rather than scale individual AI use cases in isolation.

quote · high
…era of chatbots feels ancient,” says Olof in the AI After Work (AIAW) podcast. Instead of scaling individual AI use cases in isolation, Nordea chose to build a shared, flexible platform leveraging Amazon Web Services (AWS) Bedrock that works with different models.…

Solution

Nordea built a shared, flexible platform leveraging Amazon Web Services (AWS) Bedrock that works with different models, avoiding vendor lock-in. Governance, including monitoring, access controls and compliance, is incorporated into every layer, and security, privacy, cloud, risk and product teams collaborated from day one.

quote · high
…magic solution, it was deliberately framed as a systems‑engineering challenge. It required security teams, privacy experts, cloud architects, risk officers and product owners to collaborate from day one,” Olof adds. This way of framing and approaching AI succeeded. Embedding AI acro…

Business value

The source states that scaling to more than 10,000 users moved the work from laptop pilots to a production-grade platform; it gives no efficiency figures.

quote · high
…prototypes. Scaling to more than 10,000 users required discipline and patience. The work moved from laptop pilots to a production‑grade platform. It was delivered in stages. “Sometimes a release is a breeze, s…

AI capabilities

Generative AI

classification · high
…leveraging Amazon Web Services (AWS) Bedrock that works with different models. This is a managed service from Amazon that provides access to a variety of large language models. This strategic choice avoided vendor lock‑in and enabled flexibility as AI mode…

Implementation approach

The platform was delivered in stages, split into multiple parts so that partial delivery still went to production; the focus is now on transforming workflows rather than building prototypes.

quote · high
…was delivered in stages. “Sometimes a release is a breeze, sometimes you sweat. We tried to split it into multiple parts so that if we only got 80%, that 80% still went to production,” Mattias notes. The experience shows a few simple lessons for leaders: Start sm…

Headline outcome

derived · high
…ea, the focus is now on transforming workflows rather than building prototypes. Scaling to more than 10,000 users required discipline and patience. The work moved from laptop pilots to a production‑grade platform. It was delive…
Capture details
Captured
14 Nov 2026, 06:02 UTC
Extractor
fetch-strip@1
Snapshot hash
f36af11f34bc1177607840aa05d85532e3595c31727f67058168cad325915975