March 28, 2026 · 13 min read
Every Company Is a Talent Management Company
By Ayush Kumar
In the summer of 1950, an agent named Lew Wasserman walked into Universal Pictures with a proposal so strange that the executives initially thought it was a joke. His client, Jimmy Stewart, had signed to star in a mid-budget Western called Winchester '73. Stewart's quoted fee was $200,000, which Universal didn't want to spend on a film of that scale. Wasserman's counter was that Stewart would take no upfront salary. None. In exchange, Stewart would take fifty percent of the film's net profits.
Universal agreed because the alternative was no Jimmy Stewart. The film became a hit. Stewart walked away with around $600,000 — three times his original quote. The studio paid more than it would have paid in salary, but it had no choice; it had agreed to the structure when it had leverage and discovered the cost only after the talent had won.
That single deal — one Western, one agent, one star — broke the studio system that had governed Hollywood for thirty years. Before Wasserman, studios owned talent. Stars were signed to seven-year contracts at fixed salaries, suspended without pay if they refused roles, loaned out to other studios the way you loan out a forklift. The studio was the asset. The actor was a worker. By 1962, MCA — Wasserman's agency — had become so powerful that the Justice Department forced it to choose between representing talent and producing films. It couldn't be both, because by then, talent had become more valuable than the production infrastructure that depended on it.
The hard part stopped being making movies. The hard part became finding, signing, developing, and retaining the people without whom the movies were just blank reels.
This is the shift that is now happening to every company.
The first inversion
For most of the industrial age, companies owned their inputs. Capital, machines, distribution, brand, and — implicitly — people. The employment contract was a kind of ownership: in exchange for stable wages and a desk, you traded your output. Your judgment, such as it was, belonged to the company. Your career was the company's to design. IBM, GM, GE — the canonical mid-century corporation — was an arrangement for converting human time into shareholder value at scale, and the people inside it were component parts of that machine, valuable mostly for being reliably interchangeable.
The studio system was just the entertainment industry's local version of the same arrangement. Louis Mayer at MGM didn't think of Clark Gable as a free agent any more than Henry Ford thought of an assembly line worker that way. Gable was a contracted asset, and the studio's competitive advantage was the catalog of contracts it had locked up.
Then came the inversion. The 1948 Paramount Decree broke the studio system's vertical integration. Wasserman invented the percentage deal. Television fragmented the market. The bargaining power of individual talent — actors, directors, eventually writers — became enormous, because they were the one input the studios couldn't substitute. Anyone could buy a camera. Nobody else was Jimmy Stewart.
The studios responded the only way they could. They stopped trying to own talent and started renting it. The structure that emerged — small agencies, finite engagements, upside ownership, careers as portfolios of projects — is what every knowledge company is now becoming. The comparison is more literal than people realize.
The McKinsey parallel
A different version of the same shift was happening in the white-collar world, decades before the rest of the economy caught up.
In 1939, a Chicago lawyer named Marvin Bower took over a struggling accounting firm and rewrote it as something new. McKinsey & Company, he insisted, would not be a "business." It would be a "professional firm" — built on the model of law and medical practices. Its product would not be services in the usual sense. Its product would be judgment.
Bower's bet was that large corporations had grown too complex for the people inside them to see clearly. The CEO of General Motors could not understand his own company's manufacturing inefficiencies, distribution gaps, organizational pathologies. He needed someone from outside — someone trained to look at a system and tell him what was broken. That someone would not stay. They would come for three months, assess, recommend, leave. The work was discrete. The judgment was the deliverable.
What Bower invented was a way to monetize judgment in pure form. Not paired with execution. Not embedded in a product. Just thinking, sold by the hour, by a person trained at Harvard Business School, deployed for a fee no salaried employee would ever command. The McKinsey consultant was the corporate world's version of the Hollywood actor — someone whose value lived in their head, who could be rented but not owned, whose career was a portfolio of engagements rather than a tenure inside a single company.
Most corporations dismissed this as an accessory. Useful in moments of crisis, mostly irrelevant. But the model kept growing. By the 1990s, McKinsey, Bain, and BCG together advised most of the Fortune 500, generated tens of billions of dollars in revenue, and supplied a meaningful share of America's CEOs. The consultants weren't replacing the company's employees. They were doing something more interesting and more telling: they were performing the function of judgment, on demand, for a company whose internal structure had grown too rigid to produce it on its own.
What McKinsey understood, before anyone else, was that the corporate form is structurally bad at developing the one thing it most needs.
The compression
Sixty years later, that pathology has metastasized.
Jack Dorsey cut roughly four thousand people at Block in February 2026 — about forty percent of the workforce. The cuts targeted middle management almost specifically. Dorsey's argument, published with Roelof Botha a month later, was that hierarchy existed to route information. AI now routes information better. Therefore, the function performed by middle management is structurally obsolete, and the people performing it are too. Meta's AI team had already moved to a 50-to-1 engineer-to-manager ratio. Amazon flattened reporting structures the same year. Vinod Khosla had been saying for years that 80% of jobs would be done by AI by 2030 and $15 trillion of U.S. labor GDP would mostly go away.
What survives this compression is not the people who route information or coordinate execution. What survives is the people who decide. ICs build, but they decide what to build. DRIs own outcomes, but they decide which outcomes matter. Player-coaches develop people, but they decide how the organism grows. Every surviving role is a judgment role. Every other role is being automated, eliminated, or absorbed into the intelligence layer.
This means something specific that most of the commentary has missed. When a company strips out everything except its judgment roles, what remains is a structure that is no longer recognizable as a corporation in the old sense. It is recognizable, instead, as a talent agency.
The remaining people are not "employees" in the way that word used to mean something. They are not interchangeable holders of standardized positions. They are individuals whose specific judgment is the entire reason the company exists. The company's job is to find them, sign them, develop them, deploy them, and keep them from leaving for a competitor. The company's economic value is a function of the quality of those individuals and the structure that lets them produce their best work. That is not a manufacturing problem. That is not a process problem. That is a talent management problem.
Every company is a talent management company now. Most of them just haven't admitted it.
Every person is a consultant
The mirror image of this shift, on the individual side, is the consultant.
For fifty years, the McKinsey associate was a peculiar role — a generalist hired for judgment, deployed across problems, never embedded. Now every knowledge worker is becoming one. Not in title. In shape. The cost of an experiment dropped to a weekend. AI agents handle the execution. What you sell, what your career is actually composed of, is your ability to look at a problem you've never seen before and have a useful opinion about it within hours.
Compensation has already adjusted to reflect this. The senior AI researcher commanding nine-figure packages is not being paid for execution; ten people in the world could write the code. They are being paid for the judgment that decides what code is worth writing. The principal engineer at a frontier lab is not being paid for output; the agents produce more output than a hundred engineers could in a lifetime. They are being paid for taste. The compensation gradient inside tech, which has steepened drastically in the past three years, is the market's way of admitting that judgment is now the scarce input and almost everything else is commoditized.
This is what everyone is a founder, everyone is a VC leaves out. Everyone is also a consultant. Selling judgment, by the engagement, to whoever can use it. The free agent in the William Morris sense — the actor whose career is a portfolio of projects, who is signed to no studio, who is in continuous negotiation with the next thing.
That is the new employment contract, whether or not anyone has signed it.
The two hard problems
This shift collapses the priority list of what a company has to be good at.
There used to be many things a company had to do well. Manufacturing. Distribution. Branding. Process optimization. HR. Legal. Finance. Each was a function, each was difficult, each required dedicated infrastructure. The internal complexity of the firm was the price of operating at scale, and the management discipline was about making that complexity hum.
Most of that is now solvable. Manufacturing has been outsourced for decades. Distribution is largely AWS and Stripe. Branding is increasingly generated. Process optimization is what the agents do. Legal, finance, and HR have specialized platforms and AI tooling that compress the work by an order of magnitude every year. The functions that used to define a company's operational moat have been commoditized into infrastructure, available to any startup with an API key.
What is left, what cannot be commoditized, what remains genuinely difficult, is two problems. Capital — getting access to it, allocating it well. And talent — finding it, developing it, retaining it, directing it. Everything else flows from these.
Of the two, capital is the easier problem. Private equity, venture capital, sovereign wealth, family offices, debt markets — the supply of capital is enormous and increasingly liquid. There is more capital looking for returns than there are quality opportunities to deploy it. Capital is, at this point in the cycle, the abundant input.
Talent is the scarce one. Not warm bodies — those exist in surplus. The specific cognitive and aesthetic capability the work requires: judgment under uncertainty, taste under ambiguity, conviction under pressure. The people who can look at a market and see what is worth building. The people who can read a customer call and know what is actually being said. The people who can manage a hundred agents without losing the thread of why any of it matters.
These people are rare, they know they are rare, and they have unprecedented optionality. They will not be retained by titles or processes designed in 1985. They will be retained, if at all, by structures that look much more like talent agencies than corporations: small teams, finite engagements, ownership of upside, autonomy over how the work gets done.
The new corporate form
The companies that figure this out first will look strange to the companies that don't. They will not have many layers. They will not have HR in the traditional sense. They will not have career ladders, performance review cycles, headcount targets, or any of the apparatus that constituted corporate adulthood in the previous era. They will look like ensembles. A small core of permanent talent, a rotating roster of specialists on finite missions, a substrate of AI agents executing under direction. They will pay obscenely for the people they need and they will let everyone else go without ceremony.
The structure is not new. It is the William Morris structure, the McKinsey structure, the Hollywood talent agency structure, ported into the operating company. The moat is not the brand or the product — both are increasingly commoditized — but the roster. Who you have working for you. What they can do. Why they will not leave. Talent becomes the key result, in the literal sense: it is the output the company exists to produce, manage, and compound.
This is the part that is hardest to internalize for incumbents. A car company exists, in some final sense, to produce cars. A bank exists to allocate capital. An ad agency exists to make ads. The talent inside each was historically a means to that end. What the inversion suggests is that the relationship reverses. The car company increasingly exists to assemble and deploy automotive judgment. The bank increasingly exists to assemble and deploy financial judgment. The ad agency increasingly exists to assemble and deploy taste. The output is downstream of the talent. The talent is the company.
What this means for hiring
If talent is the central asset and judgment is the central capability, then hiring is the central function. Not as a back-office HR process. As the most strategically important thing the company does.
Hiring as it currently exists is structurally unfit for this. Resumes optimize for credentials, which proxy poorly for judgment. Interviews test verbal fluency under stress, which is correlated with neither competence nor character. Reference checks are theater. The entire apparatus was designed to fill specialized positions in a hierarchical structure where the cost of a wrong hire was bounded by the role's narrowness — a mediocre middle manager could only do so much damage in a system where her decisions were filtered up and down a chain.
That damping function is gone. In a flat organization where every IC owns outcomes and directs agents, a person with bad judgment is not corrected by their manager — there is no manager. Her decisions propagate directly into the work. The cost of a bad hire is no longer a quarter of dragged-down performance; it is a year of compounded mis-direction across a hundred agents. The cost of a great hire is a step-change in what the organism can do. Hiring moves from administrative process to existential bet.
Everything that screens for judgment becomes valuable. Everything that doesn't becomes a liability. The companies that figure out how to test for the actual cognitive profile they need — the way McKinsey figured out how to test for it with Solve, the way Hollywood figured out how to test for it with table reads and screen tests — will compound. The ones still optimizing for resume keywords will hire smart-sounding people who cannot make consequential calls, and will wonder why their flat org keeps failing.
The closing turn
The economic logic of the next decade is straightforward to describe and hard to live inside. AI eats execution. Judgment becomes the only scarce input. Capital floods toward whoever has the judgment to deploy it. Companies compress into the people who can produce judgment and the agents that execute on it. The corporate form looks more like a talent agency than a factory. Every person inside it looks more like a consultant than an employee. The relationship is mission-based, upside-shared, terminable on short notice in both directions.
Some of this is exciting. Some of it is brutal. It is, like all the previous shifts — the printing press, the assembly line, the platform, the algorithm — neither good nor bad in itself, just a redistribution of where the value lives and therefore where the leverage lies. The people who recognize it early will design their lives accordingly. The companies that recognize it early will design their structures accordingly. The ones that don't will keep running the 1985 playbook and wondering why the people they want will not stay.
Lew Wasserman did not invent the talent agency. He just recognized that the existing arrangement — studios owning stars — was about to invert, and built the structure that worked on the other side. The studios were not stupid. They had simply optimized for the previous game.
Every company is a talent management company now. Most of them are still optimizing for the previous game. That is the opportunity, and it is also the warning.