Lean is dead: the future of work is adaptability

Start with an old African proverb. Every morning in Africa a gazelle wakes up and knows it must outrun the fastest cheetah or it will be killed. Every morning a cheetah wakes up and knows it must run faster than the slowest gazelle or it will starve. It doesn't matter whether you are a cheetah or a gazelle. When the sun comes up, you'd both better be running. Evolution has matched the two of them rather well.

Then a third species walks onto the plain: creative humans, in camouflage, with guns. The old rules of evolution do not apply to this sudden change. The new predator has appeared from nowhere with what look like magical skills, and slow Darwinian evolution cannot equip the cheetah or the gazelle with a defence in time. For each human visible, ten more are over the horizon. Running fast won't outpace a bullet. Agility might dodge one. The prey will only survive by adjusting their existing senses and reacting more quickly. Learn the sound of cars. Learn the smell of humans. Learn the new patterns that didn't exist a generation ago. Be willing to change direction more often.

This is what has happened to lean. Being lean is wonderful when the environment is stable, when the predators you face today are the ones you faced last year. Costs come down. Processes get sharper. The organisation hums. The problem is that we do not live in a stable environment. We never quite did, but we used to be able to pretend. We can't any more. Lean as the default is the cheetah refusing to learn the sound of cars.

That was the thesis I took to a Future of Work conference in late 2019. The talk was called Lean is dead! The future of work is adaptability. I called the model I proposed “Orga*gile 4.0”, a deliberately clumsy compound to make the point that organisations needed their next operating-system upgrade, after BPR, after Total Quality, after Lean. Six years on, after the pandemic and in the middle of AI disruption, the name feels dated but the argument has held up better than I expected. This is a revised version.

The opening, almost verbatim from 2019

What lean got right, and where it ran out of road

Three decades of efficiency and effectiveness programmes (Business Process Re-engineering, Total Quality, Lean) produced very capable organisations. Better unit economics, less wasted effort, cleaner processes, fewer surprises. There is nothing wrong with any of that. I am not arguing for waste. I am arguing for enough slack that people are not running at full capacity all the time, because a team working at its limit has no time left for development or innovation.

The problem is that lean delivers its benefits under predictable change. If you can forecast the next quarter, the next year and the next product cycle, lean is the right tool. You compress the variance out of operations and let the predictable flow run faster and cheaper.

The regime we actually live in is one of structurally unpredictable change. Connectivity, distributed knowledge, AI, geopolitical reshaping, climate response, demographic shifts, supply-chain shocks. None of these is background noise. Each is a structural force. And the response lean teaches (tighten, optimise, eliminate waste) actively makes an organisation worse at coping with structural force. Slack is not waste in this regime. It is the budget for sensing and responding.

What changed, and what we missed

In the talk I borrowed a line from Professor Eddie Obeng: there has been a step-change in the world, and we all missed it. The change is the move from a world where most problems are complicated (more analysis solves them) to one where most strategic problems are complex (no amount of analysis solves them, because the system is responding to itself faster than you can study it). Lean is wonderful for complicated problems. It is the wrong tool for complex ones, and increasingly the complex problems are the only kind worth a senior leader's time. Cynefin is the cleaner version of this argument. In the talk I made it without the vocabulary.

The forces I named in 2019 were the standard five: complexity, pace, ambiguity, volatility, uncertainty. If I gave the talk now I would add a sixth: AI as a general-purpose accelerant. It does not replace the others. It multiplies them. A complex problem evolves faster. An ambiguous signal gets amplified. A volatile market moves before the response cycle completes. We have not yet adjusted our organisational reflexes for that.

The upgrade: three things, deployed together

What I argued in 2019, and still argue, is that adaptability is not one thing. It is three things deployed together, inside an outer ring of trust and technology.

Agility at the core is the capacity to act at speed once the response is chosen. Many businesses think this is the part they need to invest in. It is usually the easiest of the three.

Collaboration, collaborative mindset and co-creation make up the middle ring. I covered these in detail in the Conscious Collaboration whitepaper a few months after this talk. I put three things in one ring because they are different muscles. Collaboration is the structural process. Collaborative mindset is the behaviour and the language. Co-creation is what happens when both mature enough that work is genuinely shared rather than just consulted on. Many businesses have one of the three and assume they have the others.

Trust and technology form the outer ring. They are the enabling infrastructure: the platforms that let distributed teams work, and the trust that makes those platforms productive instead of theatrical. Trust without the tools is a dinner party. Tools without the trust are a surveillance system. Together, they are the operating environment for the three disciplines in the middle ring.

Scan, connect, prioritise, respond

The mechanism underneath all of this is a four-step loop that runs continuously. Scan the wider environment. Connect the signal to people who can act on it. Prioritise using whatever decision framework fits the type of problem (Cynefin works well here). Respond with the appropriate kind of Idea Mining or execution. Archive the result, then start again.

None of this is fast in absolute terms. The speed comes from running the loop often, not from any single cycle being short. The lean organisation runs the loop once a year, calls it strategy, and goes back to optimising. The adaptable organisation runs it continuously and treats strategy as the moving aggregate of those cycles.

What six years have done to the argument

Three observations from 2026.

The pandemic was the rehearsal. The organisations that adapted in 2020 were already structurally adaptable. They had distributed collaboration, decision rights pushed out, and technology that supported both. The organisations that struggled were the ones that had spent the previous decade compressing slack out of their operations in the name of efficiency. An unwanted natural experiment vindicated the argument.

AI has made cognition cheap. The forces I named in 2019 have not changed in kind, but the floor under the response loop has dropped. Scanning, connecting, prioritising and even responding can now happen at speeds that were unthinkable when I gave the talk. The organisations that internalise this are not the ones with the biggest AI budgets. They are the ones whose collaboration, mindset and trust are mature enough to absorb the new speed without breaking.

Lean isn't actually dead. I gave the talk a deliberately sharp title to make people pay attention. The honest version is more nuanced. Lean is the right discipline for the parts of an organisation that operate in stable, complicated regimes: supply chain, financial controls, regulated processes. It is the wrong discipline for the parts that face a complex, fast-moving outside world: strategy, product, innovation, partnerships. The fully adaptable organisation has both. The trick is knowing which is which.

Operating across the boundary

As a CTO, the application is straightforward. I look for the parts of the technology organisation that should be lean: platform reliability, change management, incident response, cost discipline. I look for the parts that should be adaptable: product strategy, architecture choices, where to spend the next year of R&D. I make sure those two halves are not run by the same playbook, and that the people running them know which half they are in.

The hardest move is at the boundary. Lean operations want to absorb the adaptable parts because they look chaotic. The adaptable parts want to be left alone because lean processes feel slow. A leader's job is to keep both alive in the same organisation, and to know when to move things across the boundary as conditions change.

That is what I meant in 2019 by “the future of work”. It is what I still mean now.

Four things worth taking seriously

For boards: in an unpredictable environment, slack is not waste. Before approving the next efficiency programme, ask what sensing and responding capacity it removes.

For CTOs and engineering leaders: decide which parts of the organisation should be lean and which should be adaptable, run them on different playbooks, and make sure the people in each know which half they are in.

For founders: run the scan, connect, prioritise, respond loop often. Strategy is the moving aggregate of those cycles, not an annual event.

For anyone investing in AI: AI multiplies the forces lean struggles with. The return depends less on the size of the AI budget than on whether collaboration, mindset and trust are mature enough to absorb the speed.

I would be interested to hear which parts of your organisation are running on the wrong playbook: lean where they need to adapt, or adaptive where they need to be lean.

Related: Conscious Collaboration · Cynefin and Conscious Collaboration · Idea Mining · all writing