[00:00:00] The interface is leaving the desk wearables, voice AI agents in your ear. Most workers won't need a desk and a keyboard. The entire infrastructure of office work, the desks, the chairs, the monitors, the parking lots, the queue at the cafe, the morning traffic, was built around an interface we're about to leave behind. When AI handles the work that drained us, humans get to do the work we're actually good at.
[00:00:25] Welcome to the Agency, our podcast. We have a very different episode for you this week. I recently, and I mean recently in the last couple of days, published a manifesto called Headless the Future of Work. You can find
[email protected] for those of you that prefer audio. I'm just going to read it to you word for word. Here we go.
[00:00:44] There's a building in every capital city that used to make sense. It doesn't anymore. It's full of people who'd rather be somewhere else doing work a machine could do in a building they're paying to be inside on someone else's schedule. You in a job they took because there wasn't a better option. The lights go on at 7. The lifts fill at 8:30 by 9 o', clock, every floor is humming with the sound of humans typing words into screens that will be read by other humans who will type back on platforms that exist because nobody invented anything. Better yet, we built a civilization around that arrangement and we called it success.
[00:01:19] We're done. This is for the agency founder doing 800 grand with 12 employees and the executive at the Fortune 500 with 12,000.
[00:01:27] The disease is the same and the cure is the same. You're running out of life and you've been spending it inside a model that stopped making sense about 18 months ago. I'm 52 and I've watched friends die at 46. At 52. At 56, they left behind young kids, half built businesses, plans they never got around to. We're not here to fuck spiders, so here's where we are. Headcount is not a flex. Somewhere along the way, how many people work for you became the proxy for how successful are you? And that question is broken. A 50 person agency at 5 million with a burnt out founder is not winning. A 12,000 person corporation where 8,000 of those people sit in meetings about meetings is not winning. A solo operator at a million dollars with their weekends back is not losing. We've been measuring the wrong thing for so long, we forgot we picked the metric. Look at how founders introduce themselves at events. We're 47 people now. We just crossed 200. We've got three offices in three cities. Nobody says our margin is 78% or I work 12 hours a week, or we ship the same output we shipped two years ago with a third of the staff.
[00:02:34] Those are the numbers that matter. They're also the numbers nobody brags about because we've trained ourselves to read them as somehow less serious. Every Headcount announcement is a press release for a problem you couldn't solve. Every we're hiring post is a confession that whatever system you've built doesn't work without more humans inside it. Hiring is what you do when you've run out of ideas about how to do it differently.
[00:02:56] Payroll is the most expensive form of procrastination ever invented. The size of your team is the size of the bet you've made against your own Systems. And for 50 years, that was a good bet. The systems weren't there. Humans were all you had. If you wanted more output, you hired more humans and you accepted the tax that came with them. Hr, payroll, recruitment, training, performance reviews, exit interviews, the org chart, the Christmas party, the lawsuit risk, the 30% who quit every two years and take their training with them. Headcount was the price of doing business at scale. Twelve months ago, that stopped being true. If your business needs 50 people to run, you don't have a business. You have a job that 50 other people also have, and you're the one who has to make payroll. Hustle Culture was sponsored content. Rise and Grind was a sales pitch dressed up in motivational quotes by people who needed you to work harder so they could get richer. The men who told you to sleep when you're dead were either lying about how much they slept or. Or actively destroying themselves on camera as content. Half of them are dead now, the other half are in therapy. The deeper sin. Hustle culture made suffering itself the proof of seriousness. If you weren't exhausted, you weren't trying. If you weren't burnt out, you weren't ambitious. If you weren't sacrificing your kids, school plays, your wife's birthday, your own health. You didn't really want it. We just couldn't see it because everyone around us was nodding. And the wildest part is how many of us bought it.
[00:04:19] Smart people, thoughtful people. People who would never tolerate this logic in any other domain. If a personal trainer told you that the only way to get fit was to never sleep and resent everyone you loved, you'd find a different trainer. Somehow, in business, it sounded like leadership. Burnout is a receipt for time you don't get back. You can't expense it, you can't bill for it, and you can't tell your kids when they grow up that the reason you miss their childhood is that you wanted to be the kind of person who who miss their childhood. I have a young family. I don't want to work 60 hours a week. I don't want to be the guy at the school pickup who's there in body and gone in his head mentally running a Q3 forecast while his daughter shows him a drawing of a dog. I've watched too many people I respected do that for 30 years and end up with money, a stranger's marriage and kids who don't call. The brag of the next decade is the unhurried one. Anyone still telling you to grind is selling you something. The thing they're selling you is your own life. Back to you in installments. SOPs and VAs were a workaround. Standard operating procedures and virtual assistants. The workaround is over. A generation of business owners read the four hour workweek and built their entire operating philosophy on it. Document everything. Hire cheap labour in emerging economies. Train them up. Delegate yourself out of the business. Live free. It almost never worked. Every founder I know who tried it has the same stories. The VA who quit after six months and took the SOPs with them. The team in the Philippines that need 4am calls and daily hand holding. The loom videos that no one watches. The the notion docks that became graveyards. The trello boards that filled up with cards. Nobody moved. The endless cycle of hire train hope replace hire train hope replace that 8 more founder time than just doing the fucking work yourself. I've watched it run for 15 years. I've coached agency owners through every version of it. I've tried every version of it myself. The SOPs we wrote were documentation of a system that didn't exist yet. They were hope dressed up AS process.
[00:06:12] The VAs we hired were humans. We'd train to do work that could be done by software. Software that doesn't quit, doesn't sleep, doesn't take its training with it when it leaves, doesn't have a bad week because its cat died. Doesn't ghost you for a slightly better offer in month seven. Tim Ferriss wrote a good book and that book was right for 2007. We've been running its playbook for 20 years because nobody had a better one. Well, now we do. You don't need to document a single SOP ever again. You need a skill MD file and an AI that reads it, the AI follows it, consistently. Doesn't argue about it, doesn't lose the link, doesn't quit, doesn't take your IP with it when it leaves. You spent 15 years trying to engineer humans into machines, or the machines have arrived. They're better at being machines than the humans ever were. And this is the part that most business owners still don't want to hear. The thing you've been trying to build for two decades, the leveraged business you read about in 2007, it's finally possible. It's just not possible the way we're doing it. The humans you spent years training. The SOPs you spent months riding, the offshore team you spent thousands recruiting. All of it was infrastructure for a problem that just got solved at a different layer of the stack. And that's not your fault. You executed a 2007 strategy because it was the only strategy available. The world changed when you weren't looking. Now you get to choose. Keep running the workaround or build the actual thing. So here's what headless means. We need to define this properly, because the word is doing a lot of work in this manifesto, and it didn't come from nowhere. In software, headless means the back end, where the logic and data live is separated from the front end, the screen that humans look at. The head is the ui. Headless software doesn't need a UI to do its job. APIs and protocols let any front end connect to the same backend. A web page, a mobile app, a smartwatch, a Voice Assistant, an AI agent. For 25 years, using software meant a human in a browser, clicking buttons, filling forms, reading menus, squinting at dropdowns. Every SaaS company on Earth spent billions designing screens that humans would look at for eight hours a day because humans were the only entity that could read the screen and decide what to do next. And that's the assumption that just died. AI agents don't need to know where the button is. They don't navigate, they don't scroll, they don't squint. They call the function directly. The UI becomes one of many ways to access the system, and increasingly the slowest one. This is already published company strategy at every major software company on the planet right now with while you read this, Salesforce announced Headless 360 at TDX 2026. Every layer of the platform is now open to AI agents through APIs and Model Context Protocol servers. They shipped 60 new MCP tools at launch. Salesforce co founder Parker Harris said it from the stage. Headless doesn't mean no ui. It means decoupling capabilities from the interface. Meta shipped its first party Meta Ads MCP server in April. Google shipped Google Ads MCP. Amazon shipped Amazon Ads MCP. Anthropic created the Model Context Protocol in late 2024 and open sourced it. By December 2025, it had 97 million monthly SDK downloads and 10,000 active servers in production. It's now governed by the Linux Foundation. Founding members include Anthropic, OpenAI, Google, Microsoft, AWS, Block and PwC. When the biggest software companies on earth spend billions removing the requirement for a human to click a button, they're not making a small bet. They're telling you where the work is going in the most expensive language they have. And here's what that means for you. If software is going headless, the businesses that run on software go headless too. It's the same architectural principle, just one layer up. Separate the strategic layer from the execution layer. Let machines handle execution and let humans handle strategy. A business is software with humans inside it. When the software no longer needs humans to operate, the business doesn't either. And that's the pattern of the next decade. Headless is a unit of productivity. Now we can define it properly. A headless business decouples the strategic layer from the execution layer. The human decides what should happen. The AI makes it happen. The work gets done without humans clicking buttons inside it. The unit isn't headcount. The unit is value per human. One human producing the value of 10. That's the equation, and that's what every business of the next decade is competing to become. It works at every scale. Solo operator running a $1 million agency with no staff is headless. One human, the value of 10. A five person company doing $20 million with AI doing the work that used to take 50 people is headless. Five humans, the value of 50. A division inside a Fortune 500 that delivers the same output with a tenth of the headcount it had two years ago is headless. Whatever the number, 10 times the leverage per person. A government department processing claims with three case workers and an AI agent instead of 30 case workers in a queue is headless. A research lab where one PhD with the right tools does the literature review and synthesis that used to take A team of 15 grad students is headless. A media company shipping the daily output of 80 journalists with a newsroom of 12 is headless. The shape of the work is the same in every case. Human strategy, machine, execution, value per human as the measure. And this is what every business that Wants to survive. The next decade is going to do. Not a lifestyle play for tired founders looking to downshift. The companies that get there first will eat the lunch of the one still hiring like it's 2019. The next Fortune 500 will be smaller, faster, and worth more than the current one. And some of them will fit in a single house. Headless is the future of work. If every business that can go headless does go headless over the next next 10 years, the second order consequences are enormous. Most of them, I think, are good. Office buildings empty out. Not gradually. Quickly. Once a Fortune 500 no longer needs 8,000 of its 12,000 people in seats. One floor empties, then another, then the building, then the next building, then the block. The gleaming towers we built to warehouse humans for eight hours a day, the ones that we accepted as permanent features of the skyline, start looking like what they actually are.
[00:12:24] Infrastructure for a problem that we've just solved. And here's where it gets interesting. Those buildings sit on some of the most valuable land in every capital city on earth. Manhattan, the city of London, Tokyo, Shanghai, Sydney, Melbourne, Hong Kong, Singapore. We dedicated those parcels to office work in the 20th century because that's what cities needed to function. They don't need to function that way anymore. So tear them down. Reclaim the land. Turn it into parks, Public green space, community gardens, urban farms, outdoor markets, town squares, collaborative workshops where humans, humans actually want to gather. Cities designed around walking, eating, talking, making things together instead of commuting, sitting and suffering under fluorescent light. This isn't science fiction. It's the natural consequence of an interface dying. Nobody plants a flag and decides to demolish a skyline. The building's just empty. The leases lapse. The land gets repurposed. It happens, building by building, block by block, decision by decision. The same way the railway hotels emptied when cars arrived. The same way the factory towns hollowed out when the factories closed. The good version of that future requires us to build it on purpose. The bad version is what we got from deindustrialization. Empty space, ruined towns, and abandoned people. We can choose. We've never had this much time to see it coming. The interface is leaving the desk. Wearables voice AI agents in your ear. Most workers won't need a desk and a keyboard. The entire infrastructure of office work, the desks, the chairs, the monitors, the parking lots, the queue at the cafe, the morning traffic was built around an interface we're about to leave behind.
[00:14:00] When AI handles the work that drained us, humans get to do the work we're actually good at. Strategy, relationships, creativity, judgment. The things that our 20th century jobs were too small for us to ever get to Climate, biology, housing, education, loneliness. The kind of work we couldn't afford to do because we were too busy doing the kind of work a machine can now do for us. Knowledge workers have been lazy for the last 45 years. Not because they're lazy people, but because they spent every working hour inputting data into a computer, using a keyboard and a mouse, and then waiting for the computer to output that data in a different format. And we mistook that for thinking. It wasn't thinking. It was clerical work in a nice outfit. The actual thinking, the part that makes a knowledge worker worth anything, got pushed to the edges of the day, into the shower, into the commute, into the third whiskey at 11 o'. Clock. That changes now. For the first time in the history of office work, the work that requires a human gets to expand to fill the day instead of getting squeezed into whatever's left after the typing is done. And that's the gift. We don't know what to do with it yet, because we've spent 45 years deriving our identity from typing on keyboards. The next five years are going to be a slow, awkward, terrifying, glorious project of figuring out what humans are for. Now that the typing is over, a lot of people are scared. And that's fair. We're losing the version of work that defined us. The thing we called our job is becoming the thing the machine does, and we don't know who we are without it. And that's the real reason most people are afraid of AI, not the robots. It's the fear of being seen properly without the work we were hiding behind.
[00:15:37] So who's coming? Headless is for the operator who already knew something was wrong. The agency owner with 12 staff who keeps having the same conversation about the next hire and feeling sick about it. The executive at a Fortune 500 with 12,000 people who can see what's about to happen to her company and would rather be the architect of it rather than the casualty. The consultant who can't believe what AI can already do for him and doesn't want to wait for his peers to catch up. The accountant who knows her practice can be three times more profitable and a tenth as exhausting if she just stops doing it the way she was taught. The lawyer, the financial planner, the property adviser, the small business owner, the coo, the founder, the freelancer, the operator. The ones who looked at the standard playbook and felt their soul leave their body. If that's you. You're already one of us. You've been waiting for someone to say it out loud. If you're scared of putting yourself out there because you're afraid of rejection, this isn't going to work for you. You should be an employee. There's no shame in that. Some of the happiest people I know have jobs. But you can't run a headless business while waiting for permission from someone else. If you want a magic prompt or a downloadable template that fixes everything in 20 minutes, this isn't for you either. The thing I'm describing requires you to do actual work while the system gets trained. After that, the system runs and you get your life back. But the work is real. If your business plan is to scale to another 500 employees and another building more power to you, go do that. It's a different game and I'm not playing it. If you can see where the work is going and you'd rather be early than right, then you're in the right place. We've been here. We believe the model works and the math holds. The companies funding the infrastructure aren't slowing down. The interface is leaving the desk. The Fortune 500 is about to compress. The 50 person agency model is on its last legs. There's room at the front for the people who saw it coming. I'm 52. I'm running out of life. I'm building this in public so you can watch it happen. If I'm wrong, you'll know. If I'm right, you'll have been there from the beginning. The book is coming. The keynote is touring. The model is here. There's no membership card and no wait list. To feel like this you belong the moment you decide the old way isn't yours anymore.
[00:17:39] Headcount used to be the measure of a serious business. The new measure is value per human. We're not optimising the old world, we're building the next one. I'm Troy Dean. Let's go Headless.
[00:17:51] Hey, thanks for listening to the Agency Hour podcast. I'd love to hear your thoughts on the Headless Manifesto. Please feel free to share like subscribe. Leave us a review, get in touch with us, let us know what you think. And I look forward to keeping the conversation going and speaking with you again next week on the Agency Hour.
