Essay · Jul 2025 · 4 min read

Liquid Talent

Full-time employment is a coordination hack disguised as a natural category.

Full-time employment is a coordination hack disguised as a natural category. Companies hire people permanently because onboarding is expensive, context is hard to transfer, and trust takes time to build. Long-term employment amortizes those costs across years — it's a financial optimization, not a moral truth.

The instinct to treat employment as the default is old enough that we stopped noticing it was a choice. But if the underlying coordination costs collapse, the optimization collapses with them.

This isn't hypothetical. Handling context is exactly the kind of work AI is getting good at. Maintaining project memory, onboarding a new contributor to a complex codebase instantly, answering the "why was this decision made" question from three years ago — these are tractable problems for systems that can read and synthesize without getting tired.

If the friction of bringing someone new into a project drops toward zero, the whole argument for locking people into one company for years weakens. The question is what happens next.

Fractional Allocation

There's a pattern already visible at the top of the talent distribution: executives, advisors, independent consultants. These people already work the way everyone else might, eventually.

Instead of selling 100% of their time to a single company, they allocate attention like a portfolio. Some share on a primary engagement. Some share advising other teams on problems they've already solved. Some share in mentorship or knowledge transfer. Some share in pure exploration. The specific percentages matter less than the principle — that time is allocatable by commitment level, not just present or absent.

This works for executives because they're expensive enough that the coordination tax is already amortized. If AI reduces that tax for everyone else, the pattern can flatten down the org chart.

Gravity-Based Formation

In traditional hiring, companies build roles and fill them. In gravity-based formation, projects emerge around interesting problems and pull in the people who want to work on them. Strong ideas attract talent. Weak ideas don't.

Open source already works this way. The novel part is adding economic participation — making gravity-based teams something you can actually make a living in, not just a hobby. That's not a technical problem. It's an institutional one, and institutions move slowly.

Compound Knowledge Systems

The biggest hidden cost of turnover isn't hiring; it's lost context. When someone leaves, their reasoning, their context, their implicit understanding of why decisions were made — all of it walks out the door with them.

AI systems that capture not just what was decided but why could invert this. New contributors inherit accumulated wisdom instead of starting from scratch. Knowledge compounds across contributors rather than evaporating at each transition. The organization starts to look less like a collection of individuals and more like a persistent substrate that people move through.

Where This Doesn't Work

It's important to be honest about scope. Fluid organizations work where switching costs are dominant and where AI can actually reduce them. That's not everywhere.

Software, content, design, research, strategy — places where the work is mostly thinking and communicating, and AI can meaningfully offload execution. This is where fluid patterns look obvious in retrospect.

Physical work, deep relationship work (some sales, leadership, therapy), regulated roles where licenses don't come in fractions, tacit knowledge that refuses to be articulated — these don't adapt the same way. Some of them never will. Some of them will, but only when AI's capabilities shift in ways that are hard to predict.

The scope expands over time. But today, the constraint is real.

The Unsolved Economic Problems

Fluid talent breaks equity models. Current equity assumes lock-in — you earn a stake because you're committed for years. If five contributors each put in twenty percent of the work, how do they share ownership? Current answers don't exist. Token-based systems hint at an answer, but they also enable instant value capture, which tends to reward short-term thinking over long-term commitment. That tension isn't resolved.

There's also a risk distribution problem. Employment trades autonomy for stability. Fluid contribution trades stability for autonomy. Not everyone wants that trade, and nobody should be forced into it. Safety nets — health insurance, retirement, family support — were built around employment. They would need to be rebuilt or replaced. That rebuilding is a real political project, not a mechanism design exercise.

Who This Actually Fits

This isn't for everyone, and it shouldn't be. Some people genuinely need clear reporting structures, defined expectations, predictable schedules, and explicit permission to explore. That's not a failure of imagination. It's a different set of needs for different kinds of people.

Fluid patterns fit people who naturally work on multiple projects, get energy from variety, have strong self-direction, and value autonomy over stability. These are personality traits and life circumstances, not hierarchical achievements. Fluid talent isn't a promotion. It's a preference.

Open Questions

How much do AI tools actually reduce coordination costs in practice, measured against real benchmarks? Which work types adapt to fluid structures and which resist? Can we build economic mechanisms that make fluid contribution sustainable without collapsing into short-termism? How do laws and social norms — both of which were built around employment — adapt to work that refuses the employment shape?

These are possibilities, not predictions. The actual shape of future organizations will emerge from experiments that haven't been run yet, not from essays that anticipate them.

By Nikshep Saravanan, based in Toronto. More writing on nikshepsvn.com and X.