Migration has run continuously for three hundred thousand years and it is not stopping now.
Every person alive is descended from someone who left.
Out of East Africa and onward until they hit the Pacific. Bantu-speaking farmers moving south and east across a continent, carrying iron. Two thousand years of merchants and monks on the Silk Road, each one deciding individually that the life on the other side was worth the walk. Then Ellis Island, guest workers rebuilding postwar Europe, Gulf cities built out of imported labor, a generation of Bangalore engineers sent to Silicon Valley in exchange for a software industry.
The pattern never changes. People move toward the place where their work is worth more.
The largest unexploited arbitrage in the world economy
Take two workers with the same skills, education, and experience, and put one of them in San Francisco and the other in Paris. Their output and salaries diverge by a multiple, because one is standing somewhere with more capital, institutions, colleagues, and infrastructure.
Most opportunities get used up. A cheap flight sells out. A trader who buys a stock in one market and sells it higher in another pushes the two prices together until the gap is gone. The people who act on an opportunity are usually the ones who destroy it.
This one never gets used up. When one engineer moves from Paris to San Francisco, the opportunity for the next engineer is exactly as big as it was before.
So migration keeps happening. It happened when the gradient was arable land, then factories, and now knowledge work. People move toward the density of knowledge, and every technology that increased output per worker also increased the stakes of where the worker sits.
We are now living through the largest single jump in output per worker in the history of the category, and almost nobody has updated their model of migration to account for it.
The slope has never been steeper
Today, an engineer with a well-configured agent stack doesn’t produce 20% more. They produce work that would have required a team. The bottleneck moved off typing and onto judgment, taste, context, and deciding what should exist.
Under the old math, six weeks of visa uncertainty cost you six weeks of one person’s output. Today, for someone directing a fleet of agents, you lose six weeks of a team’s throughput, plus the compounding cost of every downstream decision they were supposed to make.
Companies who see this are paying out massive bonuses for new hires to join even a week or two ahead of time, because even more than before: time is money.
AI did not make location irrelevant
Every time I make this argument someone tells me remote work already solved it. If the work is digital and coordination is free, why does anyone need to be anywhere in particular?
Because only half of it went digital.
Money is flowing to construction of data centers, robotics labs, battery plants, manufacturing capacity. The frontier stopped being purely digital. Engineering teams who value speed of execution and overall velocity are in person five days a week.
The people who can do this work are rare, distributed across the world at random, and needed in a small number of very specific locations.
Physical AI has a physical people problem. You cannot build embodied intelligence with a disembodied workforce.
Being in the same room is a luxury good
The return-to-office fight got framed as a culture war, which was the wrong thing to watch. Almost every company wanted people back. The real question was which ones could actually pull it off. Introducing a return-to-office mandate is a luxury that only employers growing at an aggressive pace can command.
Because in-person is expensive now, and the lease is the small line item. It’s relocation packages, housing subsidies in cities that priced out normal people a decade ago, immigration counsel, tax equalization, spousal support, school placements, six months of someone’s life spent unpacking instead of building. Every company that got its team into one place paid for that, in cash, per head.
So look at who managed it. The labs and the frontier teams shipping the hard stuff are concentrated. They paid to concentrate. They treated physical density as a capital expenditure with a return, and they were right.
Agents are extraordinary at answering the question you asked but less so at telling you it was the wrong question. That correction almost never arrives through a formal channel. Momentum works the same way: it’s often driven by social dynamics and culture, which builds better in person.
As everything gets AI-fied, the moat is being human.
Judgment, trust, conviction, the ability to make a hard call with bad information and have people follow you into it. And being human is not a remote-first activity.
The companies winning right now have figured out that physical density is a competitive weapon. What they haven’t figured out is that the tax on acquiring it — the visas, logistics, compliance, and months of a person’s life — is paid in the one currency they have least of. Time.
Everyone is building infrastructure for AGI, nobody is building it for humans
Most AI produces artifacts: text, code, images, analysis, outputs that live on a screen.
At Gullie, we’re building a coordination layer for moving a human across a border. This requires the right infrastructure across the supply chain, payments, legal and compliance — so we can support all the edge cases nobody warns you about until it’s too late.
Agents do the vendor mapping, retrieval, the form-filling, the dependency tracking, the cross-referencing of rules that change quarterly. Humans do the judgment calls, because that’s where someone’s life is on the line. We partner with world-class law firms, moving companies and more to execute this at speed.
Our output is a person standing in a different country. The success condition is a change in the physical world.
AGI will help build the future, and we’re moving the people to help turn that vision into reality.
The race to deploy
a16z Speedrun reports that around 70% of their founders are foreign, and started Speedrun Global Founders, their end-to-end approach to guiding founders through visas, customs, housing, banking, and building local SF community.
Entrepreneur First pulled operations out of cities like Singapore and opened in San Francisco, because being in the same place is worth what it costs.
The slope has never been this steep. People will keep moving toward where their work is worth more, because that is the single most rational thing a person can do.
The only real question is how much time gets consumed by the logistics of the move. For most of history, the answer was: a lot. Weeks, months, and in some cases years.
We can do better than that, and we’re here to make it happen.