Your AI competitive advantage is temporary with Carsten Egeriis, CEO of Danske Bank

Egeriis (Bain, Winning with AI) — your AI competitive advantage is temporary

▶ Watch on YouTube · Bain & Company · 31:12

Carsten Egeriis is embracing AI more than your average CEO. At Danske Bank, he’s spent the last five years bringing the 155 year old bank towards the bleeding edge of AI innovation. They’ve had some impressive customer-facing successes, but he’s the first to say - they’re not done yet. … And he debunks the idea that AI is just about cost savings and productivity, advocating that AI is a tool for long term growth.

TL;DR

A 31-minute episode of Winning with AI, the Bain & Company podcast hosted by Sarah Elk (Bain) and Andrew Ng, published 20 October 2026. The guest is Carsten Egeriis, CEO of Danske Bank, Denmark’s largest bank (20,000 employees, a balance sheet larger than Danish GDP). Five claims:

1. Early and public. Danske launched an internal Danske GPT in early 2024 and reached over 75% adoption by the end of that year; twelve generative-AI tools were in use in 2025. In April (the year is not stated) it committed publicly to DKK 4.5 billion (about $700 million) of investment, roughly 50% more than 2023’s strategic-development budget, and a DKK 2 billion (about $300 million) productivity target.
2. Foundations first. The investment goes to an “AI city” architecture built with AWS and to data layers before applications, then to customer and employee journeys large enough to scale.
3. Growth, explained as cost. “There is not a single thing that we’re doing that doesn’t impact growth and revenue,” but cost terms are “much more credible” to the market for now. Advisers in the Danish retail network went from fewer than five to almost eight customer meetings on average.
4. Fast where the data is ready, stuck where it isn’t. A financial assistant in the mobile bank now takes over 30% of customer queries with a resolution rate above 80%, after eight or nine months: “It took us 25 years in Denmark to go from 600 retail branches to 50.” The bank’s many small exception processes (a lien on a mortgage, a divorce, a death) have not scaled: “the technology can clearly do it,” but “the data and the data links… are just not there yet.”
5. Advantage is short; culture lasts. A moat “today with the pace of change… is not something that’s there… you have a few months or six months or 12 months of head start.” Infrastructure and data products are harder to copy; “the biggest but also most intangible… competitive advantage is the leadership and culture.”

Key claims

Putting numbers on the table

Exact ROI on technology is hard to find, so the public commitment was “more about a mindset”: board and market buy-in, and discipline in following up, “testing, learning, failing fast, moving on, setting tangible targets.” The DKK 4.5 billion was triangulated from a bottom-up view of where the bank could scale, the pace the organisation could absorb, and board and market expectations. The board is supportive, but credibility to keep investing “comes with constantly having proof points.” Ng suggests the cost narrative is how CEOs must explain AI to investors even when they think in growth; Elk notes that in her experience cost pressure often comes from the board while the CEO is growth-minded.

Build or borrow

New models and open-source releases keep changing where building pays: “you can take that open source, you can pretty quickly put that into your own harness… and all of a sudden you can build a pitch book.” That changes “where there was advantage and where there isn’t.” As a mid-sized bank by global standards, Danske relies on close strategic partners, AWS for cloud and Infosys for engineering, and tests models and providers with them. At Davos he spoke of a 10- to 15-year payback on some investments. Ng: even an “ephemeral advantage lets you gain market share. That market share could be quite durable.” Egeriis: “Absolutely.”

Getting 20,000 people to adopt

Most employees are not AI-native, so culture takes many small things: demos that excite people, AI champions and super-users in each part of the business “that walk around with AI champion hoodies,” hackathons, a generative-AI week. Traditional training is less relevant “because the technology moves so fast.” Risk, compliance and third-party management, slow in a regulated bank, are part of the same advantage.

Jobs

He is careful about predicting which jobs will go, because it “creates… a lot of unnecessary nervousness,” but does not pretend: customer services will need fewer people as customers use the financial assistant, while the bank will need more people “working with agentic flows, working with data, working with cyber risk.” Strategic workforce planning is tied to the bank’s “AI big wins.” Ng sees a high ceiling: many more people would benefit from financial advice if it were cheaper to deliver.

A CEO’s time

He spends “probably most of my time now on both the tech, but also the leadership associated with tech,” including time with vendors, podcasts and trying the tools. His background is risk management, which taught him the “machine room of banking”. Technology also creates risk: there have been no physical bank robberies in Denmark in recent years; attacks now come through fraud and cyber.

Dynamic-capabilities reading

• digital-transforming/improving-digital-maturity — the AI-city architecture and data layers built before applications, partnerships with AWS and Infosys, and a public investment and productivity target.
• strategic-renewal/organizational-culture — culture and leadership named as the most durable advantage: champions, hackathons, demos, tone from the top.
• contextual/internal-barriers — the exception processes that stall on missing data and data links, and the long risk and compliance cycles of a regulated bank.

How it connects

• enterprise-ai-adoption — 75% adoption in a year, AI champions, public targets, and foundations before applications.
• automation-vs-augmentation — productivity reinvested in growth: more customer meetings, more advice.
• strategy — a head start of months rather than a moat, with culture as the advantage that lasts.
• Bornet (WorkLab) — the opposing view: early adopters build moats that become insurmountable.
• Krishna (IBM) — the same show, another CEO on concentrating investment.
• Kropp (BCG X) — augmentation as a compounding advantage, substitution as a temporary one.

What was actually ingested

The full episode (31:12, length_seconds: 1872) from the human-curated caption track with speaker labels. No stills: a podcast.

Linked entities and concepts

• Bain & Company — publisher; Sarah Elk co-hosts. Updated in this ingest.
• Andrew Ng — co-host. Updated in this ingest.
• Amazon Web Services — partner for the AI-city architecture. Updated in this ingest.
• enterprise-ai-adoption, automation-vs-augmentation, strategy — updated in this ingest.
• dynamic-capabilities — the three cells tagged above.

Dangling (single-source mention, deferred): Carsten Egeriis, Danske Bank (mentioned on two AWS sources without a page), Sarah Elk, Infosys.

Scope and reliability

A CEO describing his own bank’s programme on a consultancy’s podcast, with friendly hosts. The investment and savings figures are public commitments; the adoption figure, the assistant’s 30% share and 80% resolution rate, and the rise from five to eight customer meetings are his statements without method, and he notes the meetings rise is “not only AI.” The title’s claim about temporary advantage is his view and is disputed elsewhere in the wiki. Its value is a bank CEO’s account of where agentic AI scales quickly and where missing data blocks it.