Andy Bromberg and the trouble with a perfectly good calendar

On October 18, 2026, Andy Bromberg announced his billion-dollar idea: a calendar that joins a Google Meet using the correct Google account. Anyone who has spent a meeting’s first five minutes pleading with a login screen may consider this an entirely reasonable valuation. Bromberg was joking about the riches. He had, however, built the calendar.

He called it Pattern. Its announcement included timeboxes, meeting holds that clear themselves when a slot is chosen, and a horizontal desktop layout. Behind these adjustments was a particular customer: Bromberg, juggling roughly half a dozen personal and work accounts. He could describe the inconvenience precisely because he was living inside it.

There is something pleasingly specific about this chapter of his career. He is Eco’s chairman, CoinList’s co-founder and a partner at Amity Ventures. Yet the recent output includes a timer that notices when he wanders off task and a book remade for his daughter. A career associated with moving money has acquired an absorbing sideline in moving the furniture of everyday life.

Before the wallet, a room for conversation

Bromberg grew up in Massachusetts and went to Stanford to study mathematics and computer science. His early exposure to Bitcoin came through Balaji Srinivasan’s Startup Engineering class. The proposition initially sounded improbable. A few students bought some bitcoin, then started taking the machinery behind it seriously. In 2013, Bromberg helped establish the Stanford Bitcoin Group.

The group did research, built projects and took its enthusiasm to investors on Sand Hill Road. They were trying to persuade people to pay attention, rather than soliciting a cheque for a company. In his later recollection, the reception varied. Seven people were in that group; most were still working in crypto when he described it in 2018. An extracurricular interest was becoming a professional neighbourhood.

His first substantial departure from university took him into politics. He left Stanford in 2014 to build Sidewire with Tucker Bounds, whose background included Facebook and John McCain’s presidential campaign. Their project offered a place for informed people to discuss the news in public. The ambition was recognisable to anyone who had opened a social network looking for an explanation and emerged carrying an argument.

Sidewire launched in September 2015. By March 2017, it had about 800 verified political experts and another hundred in technology. Participation was restricted; the broader audience could read and like contributions. Staff selected stories and invited relevant people into discussions. Here was an attempt to improve the conversation by changing who could enter it and how much noise they could make.

That design also raised a commercial question. Would enough people return to watch a relatively small number of carefully managed discussions? The company closed in June 2017. The episode remains part of Bromberg’s public account of his career. A sensible product premise had met a difficult audience problem, and the audience problem had won.

The useful work behind the token sale

Later in 2017 came CoinList. It began as a collaboration between AngelList and Protocol Labs, then became an independent company led by Bromberg. Filecoin was an early customer. For an industry busy inventing new financial instruments, CoinList addressed the less glamorous business of making a sale function: processing transactions, checking investors and handling compliance.

This was a revealing place to stand during the initial coin offering boom. Excitement could bring projects to the door; it could not establish that a token belonged in the project. Speaking in April 2018, Bromberg described more than 900 inbound approaches and just three public sales completed on the platform. His opening assessment of the funding model was unusually plain: most projects were a poor fit.

He looked at teams, previous work, technical substance, legal arrangements and the economics of the token itself. A project needed a reason for its proposed network and incentives to exist. The platform’s job involved selection as well as distribution. Saying yes to a sale meant doing more than putting a purchase button beside a persuasive idea.

The attraction of this kind of work is easy to miss. There is little theatre in investor verification. There is considerable value in having it done. CoinList placed Bromberg close to a practical problem beneath crypto’s public excitement: how to turn interest into a transaction that an actual organisation could carry out.

Getting the technology out of the customer’s way

In October 2020, Bromberg announced that he was leaving his operating role at CoinList to become Eco’s CEO. Graham Jenkin continued leading CoinList, while Bromberg stayed connected as a special adviser. Eco’s appeal lay in a different part of the financial journey. CoinList served people seeking new tokens; Eco wanted a place in ordinary spending and saving.

Garrett Camp, Uber’s co-founder, was an early backer and founding adviser. Andreessen Horowitz later described its investment in a wallet intended to reward saving and spending, with Bromberg, Ryne Saxe and Henry Ault on the team. The pitch depended on a familiar consumer calculation: would this product offer a compelling enough experience to make someone move their money?

Bromberg’s public interviews kept returning to mainstream users. In May 2021, he discussed how rarely Eco needed to say “crypto” when explaining itself. The underlying technology could matter enormously to the company while occupying very little space in a customer’s attention. People still needed an intelligible reason to use the wallet.

The 2023 launch of Beam carried that argument into payments. Its interface aimed to make sending money straightforward while more complicated blockchain systems worked underneath. Bromberg discussed the wallet as a more cash-like experience. His present biography identifies Eco as stablecoin liquidity infrastructure and lists his role as chairman. The products and responsibilities have changed; the question of making complicated systems usable has persisted.

A year measured in small deposits of effort

On June 23, 2026, Bromberg recorded a different kind of balance: 365 consecutive days with at least one code commit. A commit could be a modest edit or a substantial feature. He explicitly resisted treating the count as a measure of quality. It recorded that he had sat down and done something.

He made a commit on the day his child was born, after his wife and baby were asleep. The detail is funny partly because it is so carefully qualified. Even a personal account of a coding streak comes with an explanation of the order of operations.

The same essay described a daily freewriting habit of at least 750 words. His preferred routines were short enough to survive a crowded day. Longer practices sometimes broke, and he said he did not punish himself for that. The point was making room for an activity he enjoyed, repeatedly, without requiring every appearance at the desk to become an event.

By then, the work itself had changed. In February, he had written that he had largely stopped handwriting code, using coding agents instead. He still distinguished engineering judgement from the physical act of entering the code. Architecture, maintenance and the ability to assess what had been built continued to matter. The daily practice could survive a change in who typed the implementation.

Four tries at the same shared document

The projects accumulated around annoyances he could name. In June 2025, he introduced interface0, a way to use different AI providers inside one interface, including shared memory across providers. Switching models had felt costly when a previous assistant already knew his preferences. He built a place where that history could travel.

Copapers followed in June 2026. He wanted people and their AI agents to collaborate on documents together. He counted it as his fourth attempt at the problem, after experiments involving Google Docs suggestions, Obsidian collaboration and artifacts inside interface0. Each earlier approach contributed something. Each had also missed something.

The new arrangement gave humans a familiar document editor and allowed agents to edit local files with their existing tools, then synchronise the results. That choice expressed a practical preference: work with capabilities the agent already handles well. The interface should help the participants meet, with less custom machinery between them.

In February 2026, Bromberg brought a related argument back to Stanford as a guest lecturer in a course on building large language model applications. Drawing on Eliyahu Goldratt’s Theory of Constraints, he urged builders to identify the part of a system limiting its results, improve it and measure again. Better output might require better context or tools. Buying access to a different model would not automatically fix the troublesome step.

October’s small releases supplied their own examples. Side Eye watches a work session and prompts its user when attention strays. Modelwatch proposes updates to model references in software repositories, leaving changes available for review. Both begin with a recurring nuisance. Both make the intervention more concrete than a resolution to be better organised next week.

The family project with a dust jacket

Some of the most revealing work has no obvious business plan. In August 2026, Bromberg described making a personalised edition of Winnie-the-Pooh for his daughter. He replaced Christopher Robin in the text and illustrations, designed the surrounding book and had it printed. The finished object involved image work, typesetting, a cover, endpapers and the less romantic requirements of preparing a book for binding.

People who saw it tended toward two explanations: either the job seemed impossibly laborious, or AI must have done it effortlessly. His account occupied the busy space between those assumptions. Automation made the project feasible. Considerable judgement remained, including how to preserve the illustrations’ character and make the pages look like a coherent book.

Earlier that summer, fatherhood had also sent him looking backwards. He researched the career of his grandfather, the mathematician Eleazer “Lazer” Bromberg, and wrote about his involvement in early computing and computational fluid dynamics. The result was a family investigation with diagrams and mathematical explanations. A relative’s name in an old technical report became an invitation to understand the work.

There is a second, public-facing expression of that interest in practical outcomes. Bromberg is a founding board member of FreeWorld, which helps formerly incarcerated people train for trucking careers and find jobs. Its founding network brought together Matt Mochary, Jason Green and Jason Wang. In this setting, access to a working system means a route into paid employment.

An invitation with fewer calendar invitations

For someone who has just built a calendar, Bromberg has a comically unceremonious approach to scheduling. In December 2025, he described telling people to call whenever they liked. He would answer if available and return the call if he missed it. He preferred conversations that could find their own length to the customary half-hour reservation.

He reproduced an exchange in which the other person kept trying to secure a time, despite repeated invitations simply to ring. The problem was social rather than technical: an available phone number somehow seemed less actionable than a calendar invitation. His short instruction was “just call me.”

Pattern, in that light, feels entirely consistent. A calendar can protect useful time and remove administrative fuss; a phone call can escape the fuss altogether. His recent projects keep testing such arrangements at close range. A meeting should open in the right account. A document should accommodate its collaborators. A book can be made for one child. There is plenty to do at that scale.

“Just call me.”Andy Bromberg / on making conversations easier