recurring p2p payments for cash app
a one-week hack week prototype that moved a decade-old idea into active development.
recurring peer-to-peer payments for cash app: send a friend the same amount on a schedule, like $20 to a roommate on the 10th for the internet bill. a decade-old idea that had never shipped.
design lead on a two-person hack week team. i found the prior design work, reshaped it into a working prototype with ai figma tooling and cursor, and co-presented the story internally.
recurring p2p payments moved into active development at cash app after the demo. two new hires, one week, a feature the product had been missing for years.
cash app is a peer-to-peer payments product. recurring payments are the kind that happen on a schedule: rent, bills, subscriptions. put the two together and you can subscribe to a person. your share of the internet, to your roommate, on the 10th, without thinking about it.
venmo and apple cash have offered this for years. inside the company it had been explored at least three separate times and deprioritized each time, not for lack of interest or customer value but because reorganizations kept splitting up the teams working on it.
we were both on square teams, less than a month into the job, with no experience of cash app code, products or design. we had one week.
build it for real, on real code, with the people who own it. a slide deck would not have changed anything.
recurring payments had been designed before, so the first job was archaeology. i used our internal ai assistant to surface old design files and, just as useful, to work out who on the cash app design and engineering teams to talk to. everyone we reached was generous with two outsiders.
- start from what existedthere were already designs for related features such as cash families. i reshaped those rather than starting a new visual language.
- ai figma pluginsused existing plugins to restructure the files, then built my own plugin with claude code for the repetitive parts of the job.
- targeted edits in codewith cursor i made changes directly in the prototype instead of round-tripping through figma for every tweak.
- one weekfrom first file to demo, alongside the engineering build.
daniel built the working prototype with claude code on a codebase he had never touched, on top of an existing feature, guided by cash app engineers who kept it pointed in the right direction. the demo sent a real recurring payment to another user.
- how quickly the right people said yes
- how much of the work already existed and just needed finding
- how little cash app context we needed once the tools could read the codebase
we recorded the demo and shared it with the cash app teams, then told the story at ai tuesday, an internal talk series.
recurring p2p payments moved into active development at cash app after a product lead saw the demo — the 57-second recording did the persuading. there is a project channel for it now.
the point of the talk was not the feature. it was that two people who did not own the product, had barely started, and had one week could move a decade-old idea because the tools now let you read the code, find the history, and build the thing instead of asking for it.