What Happens After the AI Announcement: Three Companies, Three Very Different Stories

Every week brings another press release about a company “embracing AI.” Far fewer companies tell you what happened six months later — inside the business, with the people who had to make it work. That second story is the one HR leaders actually need.

We looked at three organisations that introduced AI into their workforce for different reasons, at different scales, and with strikingly different outcomes: Ingka Group (IKEA’s largest franchisee), Klarna, and Vodafone. One got the human side right and is now living with a more complicated legacy than the headlines suggest. One got it badly wrong in public, admitted it, and course-corrected. One simply had to move tens of thousands of people through change at speed. Read together, they say more about how to lead an AI transition than any single case study could on its own.

Ingka Group (IKEA): When the chatbot’s failures become the plan

In 2021, Ingka Group rolled out an AI assistant called Billie to handle first-line customer service such as order tracking, product questions, general enquiries, around the clock, in multiple languages. Within two years, Billie was resolving around 47% of all customer contacts, roughly 3.2 million conversations, and saving the business close to €13 million.

What Ingka did next is the part worth studying. Rather than treat the remaining 53% of enquiries as a problem to automate away, the company looked closely at what those conversations actually contained. A clear pattern emerged: people weren’t just asking whether a sofa was in stock, they wanted to know whether it would work in their living room. That was latent demand for design expertise that a chatbot could never provide.

Instead of reducing headcount, Ingka reskilled around 8,500 call centre colleagues into remote interior design consultants training them in room planning, digital retail sales, and relationship-led selling. The new remote design channel generated €1.3 billion in sales in FY22, roughly 3.3% of total revenue, with a public target of reaching 10% by 2028. Ingka has continued to report against these figures every year since, which is precisely why the case has held up to scrutiny for so long.

The honest part of the story  In 2026, both Ingka Group and its franchisor, Inter IKEA, announced layoffs (around 800 and 850 roles respectively) citing declining sales, tariffs and organisational complexity, not AI. Crucially, the cuts landed in corporate and Group Functions roles, not among the 8,500 reskilled customer-facing colleagues. The reskilling commitment held; the wider promise that “IKEA doesn’t do layoffs” did not.

Best practices for HR leaders: 
Mine what your AI can’t resolve. The gaps in automation are often the clearest signal you have of where genuine human value lies.
Make reskilling numbers public and specific, then report against them annually. Vague pledges collapse under pressure; tracked ones survive it.
Be explicit, from day one, about which parts of the workforce a “people-first” AI commitment actually covers. Ingka’s commitment protected frontline colleagues, not head office (a distinction that only became visible when cuts arrived elsewhere).

Klarna: The cautionary tale everyone in fintech now quotes

In February 2024, Klarna’s CEO announced that an OpenAI-built assistant was doing the work of roughly 700 customer service agents, handling 2.3 million conversations in its first month and cutting resolution time from eleven minutes to under two. Klarna froze hiring, let headcount drop from around 5,500 to 3,500, and became the poster child for AI-driven workforce transformation. By December 2024, the company reported it had hired no human staff at all that year.

Then the quality problems that the efficiency dashboards weren’t built to catch started showing up in customer complaints. Repeat-contact rates climbed which is a strong signal that issues weren’t being genuinely resolved, only closed. By May 2025, the CEO publicly acknowledged that Klarna had “overestimated AI’s capabilities and underappreciated the human aspects of service delivery,” and the company began rehiring human agents in a flexible, remote “Uber-style” model, with people handling the complex, emotionally nuanced cases AI still can’t.

Best practices for HR leaders:

Measure what customers actually experience, not just what the tool processes. Deflection rate and resolution time look great on a dashboard while re-contact rates and complaint volumes quietly climb.
Don’t treat AI adoption as a headcount decision made once. Klarna’s error wasn’t using AI — it was moving to a near-total replacement before the technology had proven it could handle nuance and empathy at scale.
If a reversal is needed, own it publicly and quickly. Klarna’s credibility recovered, in part, because leadership named the mistake plainly rather than quietly walking it back.

Vodafone: Change management at genuinely large scale

Vodafone’s AI journey has been less dramatic but arguably more instructive for anyone managing change across a large, dispersed workforce. The company built TOBi, a customer service chatbot now handling around 45 million calls a month and resolving roughly 70% of digital enquiries, alongside SuperAgent, an AI tool that supports human agents on the calls that still need a person. Rather than positioning AI as a replacement, Vodafone paired the rollout with a structured, mandatory training programme for its customer care colleagues and invested £2 million in behavioural training so staff could keep delivering empathetic support alongside the new tools. On top of this, Vodafone is extending Microsoft 365 Copilot to 68,000 employees globally, framed explicitly around freeing people from repetitive tasks rather than cutting roles.

The results are notable less for their scale and more for what they say about follow-through: 85% of relevant staff completed the training, and first-call resolution rose to 77% as a consequence. That’s a training-completion figure most L&D teams would be proud of at any scale, let alone across tens of thousands of people.

Best practices for HR leaders:

Pair every AI rollout with a parallel, resourced training programme — not an optional module people can defer indefinitely.
Invest in the human skills the AI won’t replace (Vodafone’s behavioural training is the clearest example here) at the same time as you invest in the tool itself.
Track adoption and the downstream metric it’s meant to improve. A completion rate on its own tells you people showed up; pairing it with first-call resolution tells you whether the training actually changed how work gets done.

What ties the three stories together

None of these companies had an easy run. Even the “success” story comes with an asterisk, and the cautionary tale ends in course-correction rather than disaster. That’s probably the most useful takeaway of all for HR leaders weighing up their own AI transition: there is no version of this that is friction-free, and any internal communication promising otherwise will not survive contact with reality.

What separates the organisations that keep their people’s trust from those that lose it isn’t whether the rollout goes smoothly — it’s whether leadership treats employees as the people who’ll make the technology actually work, gives them the training and time to get there, and tells the truth, in both directions, about what changed and what didn’t.

 Sources:  This article draws on public reporting from Ingka Group’s newsroom, Reuters, Fortune, Personnel Today, Retail Dive, Microsoft Customer Stories, and Bloomberg. Figures are as publicly reported at the time of writing and may have moved on since.

Ingka Group (IKEA): When the chatbot’s failures become the plan

Klarna: The cautionary tale everyone in fintech now quotes

Vodafone: Change management at genuinely large scale

What ties the three stories together

Further reading