There is a small, absurd ceremony performed millions of times inside large companies. Someone needs access to a tool. They submit a request. The request becomes a ticket, the ticket acquires a number, and the number begins a stately tour of inboxes. A manager approves it. Finance may inspect it. An IT operator eventually performs the action. The software has behaved impeccably throughout. It has recorded every moment of waiting.
Seb Poole has seen enough of this ceremony. He has spent roughly a decade moving among engineering, product and operations, first through retail and logistics companies in Hong Kong, later through companies of his own. His present complaint is precise: much enterprise software tracks work without completing it. Modern, the San Francisco startup he co-founded with Aleks Tomovski, is designed around the opposite idea. An employee asks for something in a familiar channel; an agent understands the request, checks the rules, runs an approved workflow and records the result.
If the request is ordinary, the software should treat it as work, not correspondence. In the kingdom of enterprise tools, this is mildly radical. Entire software categories have grown rich by helping companies arrange their unfinished business into handsome queues.
“The system of record should not merely remember the work. It should help finish it.”The operating thesis behind Modern
A career in the handoffs
Poole started programming in his early teens, when building on the web involved fewer tutorials and more cheerful trespassing through view-source. Yet he followed a conventional academic path in Melbourne and studied law and commerce at Monash University. The compromise was inventive. He has said he selected subjects assessed by final exam wherever possible, leaving him able to work full time in technology while studying. Code supplied the immediacy; law and commerce supplied the institutional grammar.
That work took him through China and Hong Kong. By 2016 he was inside the startup world, holding senior technology and operations roles at fashion retailer GRANA. It was a useful vantage point. Ecommerce looks clean from the browser and gloriously untidy from the warehouse. Orders, returns, carriers, customer messages and stock systems meet at angles no slide deck would choose.
He next co-founded Olasso, a logistics and post-purchase platform. The company connected retailers to thousands of pickup and drop-off locations and dozens of carriers, while letting brands shape delivery communications. Poole argued then that logistics choices should account for their broader impact as well as price and speed. Soon after came Hulii, a marketplace for renewed electronics. The subject changed from parcels to devices, but the machinery beneath it was familiar: make fragmented systems legible, then make the next action easier.
The roundtables and the small desk
In 2020, after the circular-economy venture was sold, Poole began exchanging business ideas with James McGann. They were stuck at home during the pandemic and conducted what Poole called digital roundtables: conversations about their interests, opportunities they could see and previous experiences that had gone poorly. The pair chose a workplace problem, spent months talking to practitioners and managers, and developed Frankie Health.
Frankie began as a way for practices to take digital appointments, then evolved into a workplace platform. It raised $1.25 million in early 2021 and expanded internationally. Poole's remit was unusually broad. He led engineering, product, design and brand teams, met sales prospects and still tried to program most days. The job title was founder; the daily reality was departmental border control.
The scale arrived. Frankie served large organizations, including Disney and Uber, and in January 2023 it was acquired by Unmind. Poole stayed, leading in Asia-Pacific and helping the combined platform reach people in more than 150 countries. When he announced his departure more than five years after Frankie began, he remembered building early versions at a “comically small desk.” It is a better startup monument than a bell-ringing photograph: one desk, an unreasonable amount of work and no room to pretend the company is larger than it is.
Success did not make him allergic to beginnings. By late 2025, Poole was stepping away to start again. A founder who has experienced acquisition has many civilized options. He can advise, invest, acquire tasteful knitwear and develop opinions about coffee. Poole chose another empty document.
Software with a broom
The idea for Modern came from an irritation repeated across the companies he had built. Tools such as ServiceNow and Jira Service Management were useful records, but people still did the work behind them. A password reset is easy. Access approval is simple. Onboarding is familiar. Yet each can ricochet through several systems and several people, gathering delay like lint.
Modern's design separates reasoning from action. An AI agent can interpret the request and decide which path is relevant. The consequential steps are approved, code-based workflows derived from a company's procedures. Every action is meant to be checked against policy and written to an audit trail. It is a less cinematic picture of artificial intelligence than the all-knowing digital colleague. It is also closer to what an IT leader might permit near payroll, identity and security systems.
The company says it sees automation above 80 percent across IT requests. That is an early company claim, not a universal law, but it reveals the ambition. Poole is not trying to shave a minute from typing the ticket. He wants the repetitive majority to pass through without joining a queue at all. Modern can sit above an existing service-management system or operate as the system itself. The product also aims to learn from manual fixes and suggest what should be automated next, so coverage compounds rather than remaining frozen at implementation.
The third application
Modern entered Y Combinator's Spring 2026 batch on Poole's third application. Persistence makes a tidy moral, but timing is the more interesting one. Poole already knew the problem. Tomovski, Modern's co-founder and CTO, had spent years building developer platforms at Wix and Canva. What changed was the technical possibility: language models could interpret messy requests while deterministic workflows kept the actual execution bounded.
The pair moved to San Francisco and launched with a team size of two. The company's public language is combative. It calls Modern an AI-native alternative to the reigning service-management stack and names the incumbents it intends to displace. Under the noise sits a sober product question: can an agent be useful enough to act and constrained enough to trust?
Poole's earlier career offers a clue to how he approaches it. He is a founder who has repeatedly worked across departments rather than admiring them from the strategy layer. In a 2021 account of his day, he moved from engineering priorities to product, design, brand, sales calls and a little programming. He described himself as a “convenience fanatic” and liked an office near home. There is a philosophy here, modest but durable: reduce avoidable distance.
His off-hours reading is grander. He has recommended Nietzsche, Aristotle and Marcus Aurelius, alongside the programmer commons of Hacker News, Reddit and Lobsters. The combination is charming. Metaphysics asks what it means to exist; an internal IT queue asks why the new employee still cannot open the spreadsheet. A working founder eventually has to answer both, though only one comes with a service-level agreement.
The glamorous version of AI is intelligence. The useful version remembers the rule, completes the task and leaves the receipt.
Beginning again, usefully
Poole once closed an interview with a line about continually pursuing a life centered on core values. It could float away as founder wallpaper. His career gives it weight. He left a conventional course for the messier education of operating companies across countries. He built in consumer retail, logistics and workplace technology. He stayed through acquisition and international expansion, then returned to the beginning.
Modern remains early. It has the small team, the large target and the confident percentages common to a company still writing its first chapters. ServiceNow will not vanish because two founders find tickets inelegant. Enterprise systems persist because replacement is risky, institutions are complicated and yesterday's workaround is often tomorrow's critical infrastructure.
Yet Poole's complaint survives those caveats. Recording work is not the same as doing it. The latest generation of software has an opportunity to cross that gap, provided it can respect the rules on the way. Modern is his attempt to build the crossing: reasoning on one side, controlled action on the other, an audit trail beneath.
The deepest continuity in his story is not AI, or startups, or even software. It is the handoff. A parcel passes from retailer to carrier. A renewed phone passes from supplier to buyer. A growing company passes into an acquirer. An employee's request passes into a system. Poole keeps standing at those borders, asking why the next step takes so long.
This time, his answer is an agent with permission, a policy and a broom. The ticket may still exist somewhere in the machinery. If Modern works as intended, nobody will have to watch it age.