What H&M's AI learning day actually asked of its tech org

You've probably sat through an AI learning day that was a slide deck, a vendor demo and a Teams recording nobody opened twice. The format has become a ritual in large organisations: gather the technology group, show the tool, send the link and call it capability. H&M Group's Tech Academy ran a different day. People bring the judgement. The model takes the mundane work. Ownership sits with the teams, not with a centre of excellence waiting to hand down a direction.

Eleonore Nygards, AI Capability Partner at H&M Group, put that doctrine in a sentence most steering packs kill on contact. People want direction from the company, she said, but ownership has to live in the teams. The Academy split the room into two tracks and ran the same day live in Stockholm and in Bangalore, because that is where the tech organisation actually sits. Andreas Andersson, Area Business Manager for Digital Foundation, and Arunima Goswami, a Senior Product Designer, put their names on the same shift: time back for strategic work, and AI mapped into the process a team already runs.

Two tracks and two cities

One track was for people who write code: agents, context and the hands-on work of building with a model. The other they called Beyond Code, for technical product managers and for programme and operations people. That name is doing organisational work. If an AI programme only trains engineers, it has trained the wrong half of the meeting.

Nearly half the tech organisation sits in Bangalore. Most of the rest is in Stockholm, with smaller groups elsewhere. They ran the day twice, shifted for time zones. They did not wait for a single global town hall and a recording. The test of Nygards's sentence is not whether it appears on a slide. The test is whether a product manager in Bangalore can change a workflow on Wednesday without a centre-of-excellence ticket.

How they opened the day

They opened with an exercise that looks like a children's party until you hear what it was for. Pick an animal. Pick the strength you actually bring. Glue them into a superhero. Nygards's reading was that you show up with your superpowers, and AI automates the mundane. Andreas Andersson, Area Business Manager for Digital Foundation at H&M Group, said the opportunity is time and space for strategic thinking once the day-to-day burden drops. He used the word "superhuman" and then walked it back to "uniquely human capabilities the model cannot replicate." That walk-back is the useful part. Most learning decks say "augment" and then automate the meeting notes.

Arunima Goswami, a Senior Product Designer at H&M, named the actual gap: consistency. Some teams were already playing with tools. Others had no shared methods, no guardrails and no idea where AI adds value versus where it adds another tab. Then she said the sentence I would take into a Monday standup. There was a clear shift when teams started mapping AI into existing processes instead of treating it as something extra.

If the new board is a new artefact, it will die next to last year's map. If it is the planning board the team already runs, with one extra column, it might live.

What they did not claim

H&M is honest that it is early. A culture shift across a tech organisation of this size cannot be measured in a month. I would rather one retailer admit that than another bank publish a "400% increase in GenAI usage" with no denominator.

H&M trades in the United Kingdom and the United States. The pattern still travels to any tech organisation split across cities. Split the room so people who do not write code still have a door. Run it where the people sit. Put ownership on the team. Measure later. If you want a tighter version of that day, this is the AI skills work we run with teams who are past the slide.

What this means for L&D

A version of this that a learning lead can run in a tech organisation split across two cities is not "buy Miro." Name the two tracks in the invite so the product manager does not self-select out. Run it twice, same agenda, not a recording. Put one operator on the hook for follow-up boards. Nygards's job is capability partner, not innovation lead, and that distinction is the operating model. Keep the artefacts in the team's actual planning surface. Then wait before you brief the board on return on investment. H&M said a month is too soon. That is the adult version of an AI day: two tracks, two sites, one sentence of doctrine and a public admission that the return is not in yet.

Sources: Miro, 18 September 2026.