Read enough writing about building AI companies and you absorb an unstated premise: that the work happens in a twenty-mile radius of the San Francisco Bay, and everywhere else is a discount version of the same thing. I have never lived in that radius. I am building Gigaverse — a studio and its portfolio of AI-native companies — from Dhaka, with a hub in Wyoming, and I have stopped treating that as a fact to explain away.
Place is treated as destiny. It stopped being.
The location premium made sense when the scarce inputs were a specific network and a specific talent pool that happened to cluster in one place. A lot of that has thinned out. The models are the same models whether you call them from Palo Alto or from Dhaka. The best documentation, the best open tooling, and increasingly the best collaborators are wherever the internet is. What is left of the premium is real — capital and a certain kind of proximity — but it is a narrower moat than the mythology suggests, and the parts that matter most for a services-as-software studio are not the parts geography gates.
Why Dhaka is an asset, not an apology
Start with the thing people underrate. Bangladesh has for years ranked as the world's second-largest supplier of online labour, behind only India and ahead of the United States, in the Oxford Internet Institute's measurement of the global remote-work market.1 That is not a curiosity; it is a deep, young, English-working pool of people who already deliver technical work to clients on the other side of the planet as a matter of routine. Building a company that embeds into foreign customers' operations is, culturally, the default mode here, not a stretch.
I will not pretend the picture is uniformly rosy, because the honest version is more useful. The entry-level market is crowded and genuinely senior, judgement-heavy talent is scarcer than the headline numbers imply — which is exactly the talent a studio lives or dies on.
So the honest answer is that I do not really "hire senior" so much as compound it. The pool here is enormous at the entry and mid level and thin right at the top, which means the model that works is a small core of genuinely senior people, and then serious investment in taking sharp, hungry mid-level engineers and handing them more responsibility, faster, than they would get almost anywhere else — with agents absorbing the work that used to demand another five hires. What surprised me is how much of "senior" turns out to be judgement and ownership rather than years, and how quickly people grow into it when the leverage is there and the ceiling is not artificially low. The instinct I had to unlearn was solving a capacity problem by adding headcount; here, adding the wrong headcount is precisely how you lose the thing that makes a small team work.
What the Wyoming hub is for
The dual-hub is not hedging; it is division of labour. A US presence answers the things a foreign customer quietly needs answered before they will hand you their operations — a familiar entity to contract with, banking rails that clear, a legal surface they recognise. Dhaka is where the building happens. The point of the two hubs is that a client never has to think about the second one.
There is a neat recursion in this: setting up and running that cross-border apparatus — formation, banking, tax, compliance — is precisely what one of the Gigaverse ventures, FormBridge, exists to do for other global companies. The studio's own back office is the first customer of one of its products. That is the kind of thing you can only build well if you have felt the pain yourself.
Day to day it is less exotic than it sounds. The Wyoming side is the contracting, banking and legal surface — the entity a US customer signs with, the rails their payments clear on, the compliance posture they expect to see. The Dhaka side is where the building happens. Most of the time the two barely touch: a client interacts with the front, the work gets done at the back, and what connects them is a set of systems rather than a nightly phone call. The seam only becomes visible when something legal or financial has to move — which is exactly the seam FormBridge exists to smooth. On a good week the felt experience is that the company has no geography at all, just a front a customer trusts and a build that ships.
The agent rewrites the geography math
Here is the part that made me stop apologising. The whole services-as-software thesis is that leverage per person goes up — a small senior team, with agents doing the work that used to require a large one, delivering outcomes rather than hours. If that is true, then the calculus that once favoured a big team in an expensive city inverts. A compact senior team in Dhaka, orchestrating agents, can carry a portfolio that would once have needed four separate companies and four times the headcount. That is the bet behind running Gigabit, GigaCommerce, Top Dentistry and FormBridge off one senior team rather than four separate org charts.2
The timezone stops being a tax and starts being a feature. Dhaka and Wyoming sit almost twelve hours apart, which sounds like a coordination nightmare until you notice it means the working day barely has to stop. The team builds while US clients sleep; the agents run the seams around the clock; the client wakes to progress. The day doesn't end — it moves west.
The hard parts I won't romanticize
None of this is frictionless, and pretending otherwise would be the opposite of building in public. The genuinely hard parts are mundane and stubborn: moving money across borders cleanly, being taken seriously before anyone has seen the work, and the ordinary tax of scheduling a live conversation across half a planet. These are real, and some of them cost more than the location saves.
To be specific about the friction, because the specifics are the useful part: moving money across borders is the tax nobody warns you about — slower, more expensive and more paperwork-heavy than the elegant version of this story admits, and the single biggest reason the Wyoming side exists at all. Credibility is the second one; there is a real headwind in being taken seriously before anyone has seen the work, and the only thing that reliably beats it is the work itself, shipped and undeniable. And then there is the ordinary tax of the clock — the best real-time conversations land at the ragged edge of someone's day, and no amount of asynchronous discipline fully removes that. I would not pretend these are solved. I would say they are worth it, which is a different and more honest claim.
Building it here anyway
I am not building from Dhaka to prove a point about geography. I am building here because the inputs a modern studio actually needs — models, tooling, a deep pool of people who deliver to the world, a cost base that lets a small team take real swings — are here, and the one input that used to require a plane ticket has mostly dissolved. The interesting frontier of this whole shift is not another company in the Bay. It is what happens when the services-become-software playbook gets run from the places that supplied the services in the first place.
The honest caveat: one studio is not a proof. Capital is still concentrated, credibility is still something you earn against a headwind, and none of the hard parts above disappear because the thesis is elegant. I am making a bet, in public, that focus plus leverage plus a compounding operating model beats proximity. Ask me again in a few years. That is rather the point of writing it down now.