The Rise of the Multi-Language Human

The traditional organisation became very good at specialisation.

Deep expertise matters and will continue to matter.

But specialisation creates handoffs.

And every handoff creates distance between the original problem and the eventual solution.

Context gets lost. Assumptions become requirements. Requirements become specifications. Specifications become solutions.

And sometimes, several months later, everyone discovers that they solved the wrong problem extremely well.

Cross-domain people have always helped reduce that distance because they could translate between specialists.

AI gives them something more.

A person who understands the problem can now research unfamiliar territory, interrogate data, challenge assumptions, model a process, prototype, automate, test alternatives and iterate on work that previously required several separate functions.

That does not make them an expert in all of those disciplines.

But it gives them leverage.

AI gives the cross-domain person enough leverage to increasingly move from translator to builder.

That distinction matters.

The multi-language human does not merely help specialists understand each other.

They can increasingly participate in turning intent into implementation.

AI is making specialists extraordinarily more productive.

Yet I suspect it may simultaneously make cross-domain humans more valuable.

Because as execution becomes easier to access, more value moves towards questions such as:

The scarce skill may increasingly be less about knowing how to execute every step yourself and more about knowing what should be done, why it matters, what you don't know, and how to assemble the capabilities required to make it happen.

This is one reason I find Palantir's organisational model interesting.

Palantir describes Echos, Deltas and Devs as distinct roles that are intentionally designed to overlap.

An Echo does not simply gather requirements. Palantir expects Echos to identify the true problem, break it into workflows, align stakeholders and participate in building software or AI solutions.

Deltas bring deeper technical capability, but they too are expected to understand the mission and ensure the team is building the correct solution, not simply a technically successful one. Palantir Careers

The interesting part is not the job titles.

It is the overlap.

The model reduces the distance between understanding the problem and building the answer.

Specialists do not disappear.

The walls between them become thinner.

And there is now another intriguing step.

Palantir's AI FDE is an agent that can take natural-language intent and context, determine the appropriate operations and execute work inside Foundry — from data transformations to code and ontology changes. Palantir AI FDE

The boundaries between understanding, translating and building are becoming more fluid.

Perhaps none of this should surprise us.

We have always understood the power of complementary pairs.

What makes these pairs memorable is not similarity.

It is complementarity.

The modern organisation has always had its own versions.

Commercial and technical co-founders.

Product and engineering.

Strategist and operator.

Echo and Delta.

Complex problems rarely fit neatly inside one person's expertise, so we assemble different capabilities around them.

AI does not make that model obsolete.

It gives us a new kind of partner.

And perhaps small teams of highly complementary humans orchestrating multiple specialised AI capabilities.

The age of AI may not be the end of the great partnership.

It may be its reinvention.

Sam Altman recently acknowledged that he had expected AI to disrupt the economy much faster than it has.

His explanation was surprisingly mundane: the economy has enormous inertia. People continue using familiar tools, buying from familiar companies and working in familiar ways even when more capable technology becomes available. CTech

That suggests the bottleneck may be moving.

The question is increasingly not simply:

Can the technology do it?

It becomes:

Those questions live between disciplines.

Which is precisely where the multi-language human operates.

The multi-language human could easily become a flattering new label for the confident generalist who knows a little about everything and concludes that AI makes expertise unnecessary.

I believe almost the opposite.

The more powerful the tools become, the more important it becomes to understand where your own knowledge ends.

Something unexpected has happened to me as I have used AI more intensively.

I feel more humble.

There were subjects where I previously thought I knew quite a lot.

The ability to go deeper, challenge assumptions, explore opposing arguments and move rapidly into adjacent disciplines has repeatedly shown me how much I did not know.

AI hasn't simply given me more answers.

It has made the size of the unexplored space more visible.

That may turn out to be an important capability in itself.

AI without judgment can produce confident nonsense at extraordinary speed.

Judgment without humility can turn AI into little more than a machine for confirming what we already believed.

The multi-language human therefore does not replace specialists.

They know enough to recognise when specialist depth is required.

They understand what they know.

They are comfortable identifying what they don't.

And they can increasingly find, combine and orchestrate the capability required to close that gap — human or artificial.

This matters particularly in transformation.

Transformation rarely happens neatly inside one discipline.

It happens across the interfaces between business and technology, strategy and execution, people and process, old and new.

Those interfaces are where organisations tend to lose context, accumulate handoffs and generate friction.

So I suspect the transformation team of the future may look less like a sequence of narrowly defined functional roles and more like a network of overlapping capabilities.

Deep specialists will remain essential.

AI will make many of them dramatically more capable.

But around them we may need more people who can move across domains, understand the system, frame the right problem, mobilise specialist depth and use AI to shorten the distance between an idea and its implementation.

Not organisations full of people who can do everything.

Organisations full of people who can connect what needs to be connected.

Perhaps the organisation of the future will not need fewer specialists.

It will need fewer walls between them.

That, ultimately, may be what the multi-language human represents.

Not the end of expertise.

Not the triumph of the generalist.

But a new kind of leverage at the interfaces between expertise.

Because transformation is not about producing more transformation.

It is about improving our ability to change the system faster, with less waste, without breaking the organisation around it.