The 5% Redesigned The Work First; The 60% Got Nothing Back

Direct Source Verification: This story is aggregated from Forbes (forbes.com). Full reporting rights and copyright belong to the primary publisher.
Manu Khetan is Founder and CEO of Rolling Arrays, a LinkedIn ‘Top Voice’ Influencer and Creator of the R7 Framework.

Manu Khetan is Founder and CEO of Rolling Arrays, a LinkedIn ‘Top Voice’ Influencer and Creator of the R7 Framework.

getty​The most common thing a CEO says to me about AI this year is some version of the same sentence: We invested, and we cannot find the return.

In 2025, BCG put a number on that feeling. It surveyed 1,250 senior executives across 68 countries and nine industries. About 60% of companies reported no material value at all from AI, despite substantial investment. Around 35% are scaling. About 5% are getting value at scale. Note what is being counted: companies, not pilots.

Most boards read that and blame the models. BCG’s own reading points somewhere else. The firms getting value, it says, “go well beyond automation and incremental productivity improvements to reshape current workflows and invent new ones.”

They redesigned the work first, and brought the agent to the redesigned work.

BCG describes what the winners do. What the 60% did is my own reading, from 17 years inside enterprise workforces and from every AI budget I have watched up close, and I want it labeled as mine rather than as something the survey says. In those rooms, the pattern is consistent. They bought agents and left the work exactly as it was.

The ones who got something back did the unglamorous thing first. They took a process apart, decided which tasks should stay human, redesigned what was left and only then went shopping.

The public cases show the same order of operations. Octopus Energy has AI answering customer emails at a workload of around 250 people, with satisfaction higher than on the human-handled queue, 80 against 65. JPMorgan’s COIN took roughly 360,000 hours a year of document review off the desk, and that was nearly a decade ago.

IKEA is the one I keep coming back to. AI-powered chatbot “Billie” absorbed about 47% of inbound customer inquiries, over 3 million conversations. Billie’s savings over two years came to around 13 million euros. Then they retrained roughly 8,500 call-center workers into remote interior-design advisors, and that remote channel sold about 1.3 billion euros in the 2022 financial year.

Those two figures measure different things, a year of sales against two years of savings, so they are not a ratio. But nobody builds a billion-euro channel by shopping for agents.

Every AI budget I see falls into one of four patterns.

Redesigners are the 5%. They cut the work, not the people, and the people who stayed moved into work worth more than what left. IKEA is the archetype.

Reversers automated the moments their customers needed a human, paid for it in quality and corrected course. Klarna did this in public, which took some courage.

Experimenters are most of the 60%. Pilots everywhere, nothing at scale, a conclusion that AI is overhyped. Their skepticism is accurate about their own deployments and wrong about the frontier.

Installers take the output, and the humans simply exit. Output up, humans out. It is proof of power, not of governance, and I will come back to what it costs in Part 5.​

We run a 400-person business across six countries, and we are adopting AI as we speak, so I am doing this while also advising on it. The order was harder to follow than to describe.

Our delivery work runs through 18 roles, three ladders of six, from associate to director, and like every professional services firm, we allocated it by asking who could do it. That habit was the first thing in the way because real people sat in those roles, and we couldn’t cut them into tasks.

What unlocked it was stopping looking at the roles and starting to look at the client’s work. In a services firm, the work is defined the day the contract is signed. It lives in the scope. Read task by task, asking what kind of work each is—physical, cognitive or relational—and who should own it: execution, judgment or trust; it could be cut without anyone’s job title in the room. Cognitive execution became a candidate for agents. What stayed human became the new roles.

Mindset is the second thing in the way. The instinct is still to ask who fills the integration consultant slot, not what the work is. Hiring is where we are weakest. The new roles concentrate judgment, client relationship and accountability and need less of the configuration depth we used to interview for. The job descriptions changed faster than the interviews did. What helped more than I expected is a framework to teach, not a slogan to repeat. A grid people can apply to their own tasks gets you arguments, then adoption, slowly.

I am not going to tell you that puts us in the 5%. That is a result, and you don’t get to announce a result on the day you decide to go after it. Don’t buy an agent before you know where the work lives, and don’t hand it whatever it can take and leave the rest to roles designed for humans. Cut the work, not the people.

Whether AI is good enough is not the question, or rather, it is a question with no general answer. It gets asked task by task and only about today.

The question that separates the four patterns is different. Ask which of them you are funding. Then ask the harder version. Take the last AI investment your board approved and find the work that changed shape because of it. Not the tool that got installed. The work. If a process looks exactly as it did before, only faster in one spot, you have bought a tool and kept a workflow, and the 60% is where that lands.

And it points to a sharper money question: If the winners were not buying salary savings, what exactly should the agent dollar buy, especially if agents keep getting more expensive rather than cheaper? That is Part 4.​​​

Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?

Original Source
https://www.forbes.com/councils/forbestechcouncil/2026/11/18/the-5-redesigned-the-work-first-the-60-got-nothing-back/
Visit Forbes ↗
SHARE STORY:
𝕏 f in

Related Coverage in Business