The useful clue was sitting in the rubbish. At Dropbox, engineers gathered in a room to click through the same workflows, week after week, looking for the fresh bug that had escaped the respectable tests. The sessions were recorded. Then the recordings were discarded. Gabriel Spencer-Harper, then a software engineer, kept circling a small, impolite question: if the team had already captured how people used the product, why could those journeys not become the tests?
Software has always reserved a little comedy for its makers. A developer can reason carefully, write unit tests, add integration tests, and still watch a user discover a route through the product nobody imagined. Spencer-Harper encountered that fact at Dropbox and Opendoor. Meticulous, the company he co-founded in 2021, is his attempt to turn the joke back on the machine.
The product observes real journeys through a web application, replays relevant flows against both sides of a proposed code change, and shows the differences. Network traffic can be captured and mocked, which lets the frontend run without disturbing a live backend. The suite updates as the application changes. An engineer gets evidence about what moved without hand-writing every route or nursing a garden of flaky tests.
“What if there was a way to turn these into automated tests?”Gabriel Spencer-Harper on the discarded Dropbox recordings
The first company was a rehearsal
The origin story did not begin with a clean line from university to a fashionable market. Spencer-Harper graduated from the University of Warwick in 2016 with a degree combining Computer Science and Chinese. The day after his final exam, he wrote the first line of code for Moju, a startup built with his brother Milo. Moju was a search engine for brands and agencies trying to find Instagram influencers by topic, audience, and language.
Three months in, the brothers reached Y Combinator's final interview. They flew to San Francisco, faced four partners for ten minutes, and waited while a decision was made in minutes. YC declined. The rejection still gave Spencer-Harper confidence: a very young product had survived almost the entire filter.
Moju also supplied a less polished scene. On a visit to old university friends in Leamington Spa, Spencer-Harper sat on a bench in Jephson Gardens lamenting that no customers had signed up. He and Milo disagreed about whether they should build robots instead. It is an excellent founder anecdote because nothing in it resembles a pitch deck: two brothers, one bench, no traction, and robots waiting in the wings like an absurd emergency exit.
The geography was already teaching him something about company-building. Shortly before a university newspaper interviewed him in 2017, Spencer-Harper had relocated to China. He contrasted San Francisco's relentless concentration on technology with the wider mix of people he had met at Warwick. The comparison explained why he liked entrepreneurship in the first place. A startup forced technical work into conversation with markets, motives, and guesses about what people might want next. Moju's agency projects also made serendipity practical: one client began with a chance meeting in San Francisco. The lesson was not that every café conversation becomes revenue. It was that proximity changes the surface area of opportunity. Years later, Meticulous would keep that transatlantic shape, with engineering rooted in London and the chief executive spending substantial time close to customers and investors on the American west coast.
The startup gained customers and kept him busy coding, fixing bugs, speaking to buyers, and conducting user interviews. But the later engineering jobs supplied the problem that stayed. Dropbox and Opendoor showed him what software teams did after the clever work was finished: they tried to prove the clever work had not broken everything nearby.
Graduates from Warwick and starts building Moju; reaches YC's final interview.
Works in software engineering at Dropbox and Opendoor, where testing pain becomes a recurring observation.
Co-founds Meticulous and joins Y Combinator's S21 batch.
Announces a $4 million seed round and publishes the company's founding story.
Appears on Forbes' 30 Under 30 Europe list; Meticulous later announces a $15 million Series A.
A product made from consequences
Traditional end-to-end tests ask engineers to script what a person might do and assert what should happen. The method is valuable and irritating for the same reason: it makes somebody spell out the future. Complex applications contain more futures than a team can comfortably imagine, and every deliberate test becomes another small asset to maintain.
Meticulous changes the source material. It begins with behavior that actually happened. When code changes, selected sessions run against the old and new versions. Screenshots and runtime behavior can be compared. The system's deterministic browser work matters because a test that changes its answer without a meaningful product change becomes noise. Once developers distrust the messenger, even a correct warning joins the office wallpaper.
The ambition is larger than catching a red button that accidentally turned blue. Spencer-Harper has described a future in which a developer can see runtime information around a line of code before it reaches production: exceptions, performance effects, visual changes, and the number of users touched. The code review becomes less like reading a recipe and more like tasting the dish.
That distinction grew more useful as coding agents improved. When producing code becomes inexpensive, reviewing the flood becomes expensive. Meticulous now frames exhaustive testing as the missing confidence layer for AI-generated software. The company's public materials say its underlying engine is built on deterministic systems rather than simply asking a language model to judge another language model's work. The fashionable wrapper may be AI; the hard center is controlled replay.
The brothers and the network
Meticulous is another brother story. Gabriel leads the company as chief executive; Quentin Spencer-Harper is co-founder and chief technology officer after close to a decade at Palantir, where he worked on frontend engineering for Foundry. Their division of labor fits the product: a commercial thesis about trust wrapped around demanding browser and distributed-systems work.
The cap table reads like a compact history of modern developer tools. The seed investors included Coatue, Soma Capital, Base Case Capital, and Y Combinator. Individual backers named in company announcements include people connected to GitHub, Vercel, Segment, Adobe, Stripe, and Dropbox. Chemistry led the July 2026 Series A, with Menlo Ventures participating and more operators from Cursor, Hex, and the wider software world joining.
One investor, Ethan Kurzweil, offered a character sketch that is unusually on-brand: Gabriel and Quentin were precise and obsessive, he wrote, the sort of founders one would want catching bugs before they appeared. Kurzweil recalled an early lunch in San Francisco's Mission district. La Taqueria's line forced a change of venue. A seven-piece mariachi band arrived just as everyone sat down. The diligence continued through the brass.
The company's public hiring language makes the same operating preference visible. It asks for people who learn quickly, work from first principles, take ownership, and enjoy producing the output themselves. There is no promise of a mature playbook. The job is to discover what works, then turn the discovery into one.
“It lies at the intersection between economics, technology, psychology and future prediction.”Spencer-Harper's early description of entrepreneurship
The bottleneck beside the miracle
That old definition of entrepreneurship is a handy lens for Meticulous. Technology asks whether replay can be trustworthy. Economics asks how much engineering time disappears into tests and review. Psychology asks whether developers will believe an automated result. Future prediction asks what happens when machines produce more changes than humans can patiently inspect.
By the Series A announcement, Meticulous named engineering organizations including Notion, Dropbox, Wiz, ElevenLabs, Discord, LaunchDarkly, CoreWeave, and Wealthsimple. Forbes had placed Spencer-Harper on its 2025 30 Under 30 Europe Technology list. Current hiring material describes a transatlantic company with London engineering roots and a growing San Francisco presence.
Still, the most revealing part of the story remains smaller than the financing and more durable than the customer logos. Spencer-Harper noticed a repeated ritual, the useful information it created, and the waste that followed. He did not need to predict today's coding agents to see that testing was badly designed. The agents merely increased the pressure on the weak joint.
There is a stealable founder lesson here. Watch what capable people do repeatedly despite disliking it. Notice the artifacts they throw away. Follow the constraint that becomes more painful when everything around it accelerates. Spencer-Harper's first company chased a changing advertising market. His second grew from a room he had actually sat in.
Software will continue to break in ways its authors did not expect. Human ingenuity guarantees it; machine speed may multiply it. Meticulous is betting that the answer is not a longer checklist written by a more exhausted engineer. It is a system that remembers what users already taught the product, then makes every new line of code face the evidence.