Skip to main content
IKEA AI case study

March 2026

The 43% IKEA's Bot
Could Not Solve

IKEA might be a better AI case study than 90% of the strategy decks floating around right now.

The Numbers

IKEA rolled out an AI support bot. It handled 57% of customer inquiries. Most companies would look at that and stop. Nice. Lower support costs. Fewer people needed. Put it in a board slide and move on.

57%handled by AI bot
43%the bot could not solve

That is not the interesting part. The interesting part is the 43% the bot could not solve.

That failure exposed demand. People were not just looking for support. They wanted design help. IKEA saw the signal, built around it, reskilled employees, and turned the gap into a new revenue stream.

That is the real AI play. Not just automation. Detection.

Optimization Is Not Transformation

Most companies are still using AI to make the current machine run a little faster. Summaries. Copilots. Workflow shortcuts. Maybe some cost savings.

Fine. Useful. But that is not transformation. That is maintenance.

What Most Companies Do

Deploy AI to optimize existing processes. Summaries, copilots, workflow shortcuts. Make the current machine run a little faster.

What IKEA Did

Used AI failure data to expose unmet demand. Built a new service around the gap. Reskilled employees. Created a new revenue stream.

The bigger opportunity is using AI to show you where your current business model is blind.

What AI Actually Reveals

The bigger opportunity is using AI to show you where your current business model is blind.

  • 1Where customers want something you do not offer yet.
  • 2Where friction keeps repeating.
  • 3Where workflows break down.
  • 4Where a service should exist but doesn’t.
  • 5Where demand is showing up in disguised form.

That is where the real value is.

This is why so much AI talk feels shallow. Too many leaders still treat AI like a tool to deploy rather than a forcing function that changes how work gets done, how decisions get made, and what a company can even become.

The actual threat is not AI.
It is what your competitors become with it.

The Real Divide

A lot of incumbents still think the challenge is adoption. It is not. Everyone has access now. The real divide is between companies using AI to optimize the old model, and companies using it to find and build the next one.

That gap gets wider fast.

Optimize the Old Model
  • Bolt AI onto existing processes
  • Automate what already exists
  • Measure by cost savings
  • Same org chart, new tools
Find and Build the Next One
  • Use AI failures as demand signals
  • Redesign workflows end to end
  • Measure by new value created
  • Rethink what the company becomes

The startups get most of the attention, but that is only part of the story. The harder question is whether established companies can move with the same imagination while still dealing with governance, compliance, trust, reporting, and all the enterprise baggage. They cannot move recklessly. But they also cannot hide behind complexity forever.

That is the test. Can you use AI to rethink the business, not just speed up the existing org chart.

Beyond Software, Into Digital Labor

Because once AI moves past being an assistant and starts acting on behalf of the business, everything changes. Now you are not just adding software. You are introducing digital labor. Agents need onboarding, tuning, oversight, measurement, and clear ownership.

That is no longer just an IT issue. That starts touching operations, HR, workflow design, incentives, governance, all of it.

Why Most AI Strategies Already Feel Dated

They are trying to bolt intelligence onto processes that should have been redesigned in the first place.

The companies that win here will not be the ones with the most pilots or the loudest AI messaging. They will be the ones that use AI to uncover new demand, redesign workflows end to end, and build new forms of value before everyone else realizes the ground already moved.

Modern office space with warm lighting

AI is not just helping companies do old things cheaper. It is exposing what the next business should be.

That is the shift.

Not automation. Detection.

Donny Smith

Written by · March 31, 2026

· Founder & Executive Creative Director, Bttr.

Over the past 15 years, he has led creative teams and contributed to products used by millions of people worldwide, working with companies including Apple, Opendoor, JP Morgan, GE Aerospace, Pepsi, and Alterra Mountain Company.

LinkedIn

Share this perspective

More Insights

Perspectives on AI strategy, product design, and the decisions that separate transformation from maintenance.

What Is Temperature in AI
7 min read

What Is Temperature in AI

Temperature is a setting that controls how random a language model’s output is. At each step the model scores every word it knows, turns those scores into probabilities with a softmax, and temperature is applied inside that step: it divides each score before the probabilities are worked out, which sharpens or flattens the gap between the leading word and the rest. A Hugging Face guide to text generation describes the move as making the distribution sharper by increasing the likelihood of high probability words and decreasing the likelihood of low probability ones, and notes that as temperature approaches zero the sampling becomes equal to greedy decoding, where the model takes its single highest scoring word every time. OpenAI’s API documentation puts the range between 0 and 2, says higher values like 0.8 make the output more random while lower values like 0.2 make it more focused and deterministic, and recommends altering temperature or top_p but not both. The catch most teams miss is that a low number is not a lock. Anthropic states in its glossary that even with temperature set to 0 the results will not be fully deterministic, and identical inputs might produce different outputs across API calls. The operator rule follows: run cold when the work has one right answer, run warm when you want range, and never read a warmer setting as a smarter model.

What Is a Multimodal Model
7 min read

What Is a Multimodal Model

A multimodal model is an AI system that takes in and reasons across more than one kind of data, such as text, images, audio, and video, inside a single context. It maps each input into one shared representation, so it can relate a word, a picture, and a sound as parts of the same thing. The distinction that matters is how it is built. When Google launched Gemini in December 2023, it noted that the standard approach trained separate components for each modality and then stitched them together, and said Gemini was instead pre-trained from the start on different modalities. OpenAI made the same move with GPT-4o in May 2024, a single model trained end to end across text, vision, and audio so that every input and output runs through one neural network, which is why it answers audio in about 320 milliseconds rather than the seconds of lag in a stitched voice pipeline. The operator rule follows: a native multimodal model keeps every sense in the same context, while a vision tool bolted onto a text model reasons only over what it already turned into words.