The Scarcity That Moves

The business of advice is facing an existential shift with the emergence of technology: we are moving from an era of scarcity of knowledge (the library catalogue era) to a scarcity of assembly (the search engine era), and now, with AI, into a scarcity of meaning. This will lead to the end of the hourly billing model, the career pyramid becoming an obelisk, the inevitable rise of gig work and side hustles, and the emergence of the multicollarism.

In 2019, as a lowly Researcher, I walked into the library at the Palais des Nations in Geneva. Grand floor-to-ceiling shelves packed with books older than my grandmother, antique globes and maps whose borders had since been redrawn by wars and treaties. The kind of room that makes you feel the weight of accumulated human knowledge. The kind of room where you simultaneously feel relevant and irrelevant.

And the irony hit me almost immediately, as I sat down at a desk in that magnificent room, opened my laptop, and spent the day finding relevant cases on legal databases to validate my hypothesis.

Scattered across those shelves were extraordinary volumes - on diplomatic immunity, the law of the seas, nuclear non-proliferation. I flipped through a few. Dense, brilliant, painstakingly assembled. I thought about the scholars who had argued international law from those pages in the 1970s, drafting conventions, fighting cases, building the architecture of global governance - for whom the question which-book-has-the-answer was itself a form of expertise. Knowing where the knowledge lived was the skill, the profession. I wondered then, is knowledge quietly becoming obsolete, one search at a time?

Today, in 2026, I operate with stack of AI tools. A note-taker distils my calls into insights; an agent manages my calendar and an LLM aids my research. And I wonder the same thing. But, the more things change, the more they remain the same.

Don't get me wrong. I'm not saying AI is not revolutionary. It is! But as technology shifts, the bottlenecks shift too. What remains constant is the underlying human layer beneath the technology. Let me explain.

// Scarcity of information to scarcity of assembly

Google did not kill the library, it relocated the bottleneck. We transitioned from an era of “scarcity of information” to “scarcity of assembly”. The question stopped being “where is the answer?” and became “how do I put the answer together?”. Information became abundant, processing it did not. So, the functional skill that mattered shifted from retrieval to assembly.

During this transition, notably, the idea of knowledge transformed. What you searched for was still driven by your curiosity, your intellectual formation, your sense of what mattered. Judgement remained essential. This transition reshaped entire professions. Law firms. Consultancies. Journalism. Academia. It took nearly twenty years for most of them to fully absorb it. Some are still absorbing it.

With cutting-edge AI tools, the bottleneck has moved again, from “scarcity of assembly” to “scarcity of meaning”.

// Scarcity of assembly to scarcity of meaning

With AI tools’ superior power of cognition, the scarcity that remains - arguably the only scarcity that remains - is meaning. The functional skill that matters is shifting from retrieval to assembly to generating meaning. We are transitioning from “where is the answer” to “how do I put the answer together” to, at a fundamental level, “what is the answer”.

But, what does meaning mean? What is what? To me, meaning is a culmination of human knowledge, judgement, foresight, relationships and leadership.

When you ask an AI tool how energy is shaping geopolitics today vs when you ask what lessons Japan's response to the sekiyu shokku - the oil shock of 1973 - holds for nations navigating a dual energy and data infrastructure crisis in 2026, you get something different entirely. Asking the right questions, being able to correlate the past, present and the future – that is knowledge. This is not just glorified prompt engineering.

Knowing what this portends for a client, an industry, a regulatory regime, energy-intensive crypto mining operations and GPU-heavy data centres - industries whose entire economics are built on the assumption of cheap, abundant power - that is judgement.

Understanding that a government investing in hyperscale AI infrastructure this year is positioning for quantum sovereignty next - that is foresight.

Making tough, subjective decisions, backed by data but driven by values and instinct - that is leadership. A model can generate ten options and rank them by probability of success. It cannot absorb the consequences of being wrong.

And then there is the thing no model can replicate – relationships. The conversations, the relationships, the energy of the room where solutions are actually born. Not the deck that follows, but the moments just before and after it. The coffee that runs long, the question asked off the record that don’t make it to training data sets, the trust extended before it was earned. This trust is the hardest infrastructure to build and the most expensive to lose.

Knowledge, judgement, foresight, leadership and relationships were not rendered obsolete when online-search arrived. Their manifestation at work evolved. Similarly, now they are the only bottleneck between information and meaning.

// The economics of the new scarcity

So, how does this new scarcity of meaning impact the business of advice - consulting, law firms, policy shops, anyone who has historically sold thinking by the hour?

One won't be prescient in saying that AI has enhanced efficiency across the board. That is common knowledge. In the legal industry alone, 65% of professionals report saving between one and five hours a week using AI tools, with 12% saving six to ten hours. In consulting, a BCG experiment involving roughly 750 participants found 30–40% efficiency gains for junior analysts and 20–30% for experienced staff.

Two things become important. First, on the client-side, the hourly billing model is going the way of the library catalogue. The entire advice industry - consulting, policy, regulatory work - was built on the same premise: that thinking, and the time it takes to put that thinking onto a memo or a report, are the product. Input was the best proxy for result. AI has now disaggregated those two things.

Now, the deliverable is the thinking alone, “meaning”, the “what”. Thinking remains valuable. While the memo or the report are here to stay, but the time it takes to generate no longer a reliable proxy for the thinking.

However, before the new equilibrium is reached, the market will see tension - drastically reduced timelines will mean more pressure on resources. This could even result in declining internal and external appreciation for creativity in the standard outputs – the memos and reports. The incentive to think could go down. It will take some getting used to, to tell human creativity from AI creativity.

Second, as efficiency is enhanced, the harder questions are: in the short term, how are these organisations and professionals upskilling their teams so that as AI frees up their time, they are equipped to deliver outcomes that actually have meaning?

Long term, the problem is more concerning. Traditional organisations were built like pyramids - judgement at the top, grunt work at the bottom. That grunt work is now being automated. A major 2025 legal market report found firms have reduced the pace of associate hiring and cut the size of summer associate programmes.

This matters because the bottom was never just about productivity and grunt work. It was the training ground. It is where knowledge, judgement, foresight, leadership and relationships were developed.

So, as the bottom of the pyramid is thinning, what will the path from Associate to Partner look like? How will Associates learn to deliver meaning. And, what will promotions really mean – how will the job of a Junior Associate fundamentally differ from the Principal Associate?

// Rethinking organisation structures and professional careers

Are organisation structures and career pathways in for a fundamental redesign? They sure are being reimagined with AI agents for team members. Brian Armstrong restructured Coinbase recently around what he calls AI-native pods - small teams, sometimes a single person, folding engineering, design and product into one role. No pure managers. Five layers maximum. Player-coaches who both lead and build.

Some would argue the pyramid is being hollowed again, or becoming more like an Obelisk. Others would say it is becoming more Amoeba-like, amorphous, context dependent.

And what of the employment contract itself?

The pyramid isn’t the only structure being redesigned. The exclusive, full-time, single-employer model of linear, white-collar work - may be the next thing to shift. And policy-making globally is recognising this.

First, erstwhile employees will likely have many non-exclusive part-time jobs. Japan offers an early signal. Facing wage stagnation and an aging population, the Japanese government actively encouraged side-hustles from 2018, dismantling the legal barriers that once prevented salaried workers from moonlighting. Today, portfolio careers are normalised. Singapore and the EU too are enforcing rules on gig workers, providing them pension and injury coverage, in a way decoupling the nature of contract from the nature of protections.

Second, multicollar work will emerge. With AI making cognition abundant and hardware prestigious, a new category of technical work will take shape. The distinction between conventionally understood white-collar and blue-collar work will dissolve. This new "multi-collar" worker will lie neither in physical labour nor knowledge work alone, but in combining both. In these roles, the collars are not stacked - they are fused.

Third, professionals who emerge in the era of these trends - hollowing pyramids and portfolio careers will likely fracture into two categories. One will go deeper. These are the true specialists - the embedded professionals whose knowledge, judgement, foresight, leadership and relationships remain irreplaceable not despite the system, but within it. They are the human-in-the-loop: the ones who power the AI. These will likely stay in the “pyramid” (or obelisk). The other will go wider, in addition to going deeper. With T shaped careers - these genero-specialists will move fluidly between domains, synthesise across disciplines, and ask the right question. They will carry their knowledge, judgement, foresight, leadership and relationships and orbit the pyramid(s) - carrying "meaning" from one room to the next.

Combined, multi-collarism, non-linearism and non-exclusivism will look like a data privacy lawyer who has spent two years managing power procurement inside a data centre, and now advises a government, a climate tech fund, and a law firm - simultaneously, non-exclusively - on keeping their digital infrastructure sovereign. None of them could justify hiring her full-time. All three need her to orbit their pyramids.

The library in Geneva will still stand. The globes will still sit on their pedestals, borders still wrong in ways that matter. Albeit with machines that have all the answers, still waiting on the right question. Waiting for meaning.