On August 25, 2026, McKinsey published its latest State of AI global survey — 1,719 executives across 97 nations — and the headline number is brutal in its simplicity. After a year of record AI spending, only 6% of organizations qualify as “AI high performers,” meaning they attribute at least 5% of EBIT to their AI use. Here is the direct answer to why the other 94% are stuck: they bolted AI onto the workflow they already had instead of redesigning the workflow around it. McKinsey found that fundamentally redesigning how work flows has one of the strongest links to real profit of any factor it tested — and almost nobody has done it. Lova is what that redesign looks like in practice: a chat-first AI project management product where AI agents work as teammates, claiming bounded tasks on a shared board, moving them through explicit states, and leaving an audit trail every teammate — human or agent — can read.
Key takeaways
- McKinsey’s 2026 State of AI survey found just 6% of organizations are “AI high performers” (5%+ of EBIT from AI), and only about a third report any enterprise-level EBIT impact at all — roughly the same share as a year earlier, despite far more spending.
- The variable that separates the 6% is not a model or a budget. High performers are nearly three times as likely to have fundamentally redesigned their workflows around AI — and that redesign is one of the strongest predictors of bottom-line impact McKinsey measured.
- The individual layer is already solved. GitHub reports Copilot users are up to 55% more productive at writing code. Speed at the desk is not the bottleneck — what all that output adds up to across a team is.
- Microsoft’s 2026 data agrees on where value lives: organizational factors account for 67% of AI’s impact versus 32% for individual skill — and only 19% of firms have reached the “Frontier” zone where structure and tooling line up.
- Adoption is no longer the story. Stanford’s 2026 AI Index puts organizational AI adoption at 88%. Everyone has the tools. The 6% changed where the work lives.
Why does record AI spending produce no profit for 94% of companies?
Because spending buys capability, and capability was never the constraint. This is the quiet punchline of the 2026 survey: the tools work. Individuals are measurably faster — GitHub’s own numbers put code productivity gains as high as 55%, and most workers will tell you AI saves them real time every day. Yet McKinsey found that only about a third of organizations can trace any enterprise-level EBIT impact to AI, and that figure barely moved year over year even as deployment climbed. The money went in. The earnings needle didn’t.
The gap between “my team is faster” and “the company makes more money” has a name, and we’ve been writing about it all year: it’s the productivity paradox. Faster individuals don’t compound into a faster organization on their own, because the value of AI shows up in the seams — in how work is handed off, tracked, verified, and stitched back together. Microsoft’s 2026 Work Trend Index quantifies it bluntly: 67% of AI’s real impact traces to organizational factors versus 32% to individual skill and mindset — a better than two-to-one edge for the structure work runs on. You can buy every seat a copilot and still watch none of it reach the P&L, because the copilots are winning races the company never needed run faster.
What did the 6% of AI high performers do differently?
They redesigned the workflow instead of decorating it. This is the finding worth reading twice. McKinsey tested a long list of factors that might explain why some organizations turn AI into profit and most don’t — leadership, spend, talent, governance — and the one with one of the strongest links to EBIT was whether a company had fundamentally redesigned its workflows around what AI does well. High performers are nearly three times as likely as everyone else to have done it, and nearly three-quarters of them have, up from about half a year ago. The 6% aren’t running better models than the 94%. They rebuilt the surface the work happens on.
The distinction sounds abstract until you make it concrete, so here is the concrete version. “Bolting on” is adding an AI feature to the tool you already had: a summarize button in the ticketing app, a chat sidebar in the doc editor, an agent that drafts the status update you were going to write anyway. The workflow is unchanged; a step inside it got faster. “Redesigning” is changing the workflow itself — where work lives, who can pick it up, and how you know it’s done — so that an agent isn’t a feature bolted to a human’s process but a first-class participant in the process. That is exactly the mistake we’ve been calling out as coordination neglect: teams pour effort into making each agent more capable and almost none into the system that makes their combined output add up.
Redesign vs. bolt-on: the three-question test
McKinsey named the variable but didn’t hand you a way to tell which side of it you’re on. So here is a test you can run on any team in five minutes. If you can’t answer all three cleanly, you bolted AI on — you didn’t redesign around it.
- Where does the work live? If the answer is “in a chat thread, an inbox, and three people’s heads,” an agent can act but its work can’t be seen, claimed, or checked by anyone else. Redesigned work lives on one shared surface where every task has a state.
- Who can claim a task — and can an agent? If work is assigned by a human typing a name into a message, agents are guests. In a redesigned system an agent claims a bounded task the same way a person does, atomically, so two workers never silently do the same thing.
- How is “done” proven? If done means someone said “looks good” in a thread, you have no defense against work that looks finished but isn’t. Redesigned work has a verifiable definition of done that a card must satisfy before it can move — the same standard for a human and an agent.
Read those three questions again and notice what they describe. Work with an owner, an explicit state, and a verifiable done, all visible on one surface, is not a novel AI architecture. It’s a project board — the oldest coordination primitive we have. McKinsey’s billion-dollar finding is that the companies profiting from AI are the ones that moved their work onto a structure like this, and the ones that aren’t kept the old structure and added a chatbot to it.
Why is a shared board the redesign — not another AI feature?
Because a board changes where the work lives, and a feature doesn’t. This is the shape of Lova, and it’s worth being precise about the entity. Lova is a chat-first AI project management product built around a shared board where AI agents are first-class teammates. You steer the work in plain language — that’s the chat-first part — but every instruction resolves into a change on the board underneath: a bounded task claimed by a named owner, a status moved, a trail written. There is no separate “AI mode” and no agent sidebar bolted onto a human tool. The agent and the human operate on the same tasks, under the same rules, on the same surface.
That is the redesign McKinsey’s high performers are describing, expressed as a product instead of a slide. When an agent’s work is a card with a state rather than a message in a channel, the seams close: handoffs become claims, status becomes a query, and “done” becomes a check the card has to pass. This is why we’ve argued the real divide isn’t automating a team versus amplifying it — it’s whether your coordination layer can hold agents and humans as equals. Amplification only compounds into enterprise EBIT when the surface underneath it can see all the work at once. A summarize button can’t. A board can.
Why does this matter now, in Q3 2026?
Because the spending has already happened and the returns haven’t, and a frontier consultancy just told the market exactly why. When McKinsey — not a startup with a board to sell — reports that the single move most correlated with AI profit is redesigning the workflow, and that the profit gap held flat for a year despite record investment, that’s not a footnote. It’s the market pricing in the difference between the 6% and everyone else. The organizations that pull ahead in the back half of 2026 won’t be the ones who bought the most agents. They’ll be the ones who stopped adding AI to the old workflow and moved the work onto a surface built for how it actually runs now — a board where a task has an owner, a state, and a done that has to be earned, whether the worker who claims it is a person or an agent.
Frequently asked questions
What did McKinsey’s 2026 State of AI report find?
McKinsey’s State of AI global survey, published August 25, 2026 and based on 1,719 executives across 97 nations, found that only 6% of organizations are “AI high performers” attributing at least 5% of EBIT to AI, and that roughly a third report any enterprise-level EBIT impact — about the same share as a year earlier. The factor most strongly linked to profit was fundamentally redesigning workflows around AI, something high performers are nearly three times as likely to have done.
Why don’t individual AI productivity gains show up in company profit?
Because AI’s value shows up in coordination, not just speed. A faster individual saves time at their own desk, but enterprise profit depends on how work is handed off, tracked, and verified across a team. Microsoft’s 2026 data attributes 67% of AI’s impact to organizational factors versus 32% to individual skill — so without redesigning the surface work runs on, faster individuals don’t compound into a more profitable company.
What does “redesigning the workflow” actually mean?
It means changing where work lives, who can claim it, and how “done” is proven — not adding an AI button to an existing tool. In practice, that’s moving work onto a shared board where every task has an owner, an explicit state, and a verifiable definition of done, and where an AI agent can claim and complete a task the same way a human does. The workflow itself is rebuilt around the agent, rather than the agent bolted onto the old workflow.
What is Lova?
Lova is a chat-first AI project management product built around a shared board where AI agents work as first-class teammates. You steer the work in plain language, and every message resolves into a change on the board: a bounded task claimed by a specific owner, a status moved, a trail written. Because every action becomes a visible, recorded transition, humans and agents stay aligned on what was actually delivered — not just what was promised in a thread.
Is buying more AI agents enough to see ROI?
No. McKinsey’s 2026 data shows adoption is nearly universal while enterprise-level returns are not, and the differentiator isn’t how many agents you run but whether you’ve redesigned the workflow around them. Agents added to an unchanged process make individual steps faster without moving the bottom line; agents working on a shared board with owned, verifiable tasks change the coordination structure — which is where the profit McKinsey measured actually comes from.