Naeema · recurring P2P payments for Cash App

recurring P2P payments for Cash App

Cash App · hack week · February 2026

recurring payments in Cash App

send a friend the same amount on a schedule, like $25 every Thursday for the phone bill. we built it on Cash App's real code during a one-week hack week.

hack week build · Cash App's real code

paying $25 for the phone bill, weekly on Thursdays

roledesign lead

skillsproduct design · prototyping · design research · AI-assisted tooling

stackGlean · Figma · a Figma plugin i built with Claude Code · Cursor

teamme, an engineering fellow, and a senior product designer

timelinehack week, February 2026

the problem

Venmo and Apple Cash let people schedule payments to friends. Cash App had tried to build it at least four times in the past decade and never shipped it, because reorgs kept splitting the team.

the work

i was design lead on a small hack week team with an engineering fellow and a senior product designer. i found the earlier designs, adapted them into a prototype with AI tools, and presented the demo with the engineering fellow at AI Tuesday, Block's internal talk series.

the outcome

a Cash App product lead saw the 57-second demo, and recurring payments moved into active development with a Cash App product designer and engineers.

the demo · 57 seconds

a recurring payment, start to finish

recorded on staging: pick a friend, choose an amount and a weekly schedule, and the payment goes out on its own.

the engineering fellow built it on Cash App's real code with Claude Code, with help from Cash App engineers. the design came from the prototype below.

chapter 01the buildour rule: build on real code, with the people who own it

step 01 · find what exists

the allowance flow

the feature had been designed before. i used Glean, Block's internal AI search, to find this flow and the people who had worked on it.

an earlier allowance flow from Cash App Families, by another team

step 02 · adapt it with AI tools

a plugin, then the prototype

i wrote a Figma plugin with Claude Code that copies a set of existing screens and rewrites only the copy for a new feature.

that got the flow most of the way. i made the last small edits straight in the prototype with Cursor instead of going back to Figma.

step 03 · build in tandem·

because we used AI, we didn't work in a design-then-handoff order. we built in parallel.

01engineeringthe engineering fellow spun up a quick prototype to work out the backend on Cash App's real code.

02designmeanwhile i kept adding features and polish to the design.

03togetheri folded in what the backend work turned up, so the design matched what the code could do.

step 04 · design decisions·

01start and end datesa schedule needs to know when to begin and when to stop. both went into the design, and what the alpha couldn't support went on the roadmap.

02a reminder before each paymenta notification goes out before money moves, so people can cancel if plans change.

03cancelingupcoming payments show on the activity screen, where people already check what they've sent.

04working within the alphawe designed within Cash App's current constraints and infrastructure, so the first version didn't have to wait on new systems.

after the demo

after the demo

from demo to development

the engineering fellow posted the recording to the Cash App teams. it got 13 replies, and a product lead prioritized the feature.

our goal was an end-to-end flow and a codebase to start from, so the feature could finally get over the line.

01questions were mostly about fine-tuning, like what the date picker should look like. we gave our input.
02a Cash App product designer and engineers took it over, changed some decisions for engineering reasons, and added polish.

what i learned01

we got there in a week because the feature had been designed before.

we adapted the earlier work and used AI at each step: Glean to find it and the people behind it, Claude Code to build the Figma plugin, and the engineering fellow used it to learn a codebase they'd never worked in.

small experiments helped me pick the right tool fast. i also learned when to stop prompting: past a certain point, describing a detail takes longer than doing it by hand.

for design decisions we leaned on each other, our own taste, and the user research Cash App already had, then got an alpha out fast so we could hear from users sooner.

tools, ranked02

001Cursor edits straight in the prototype
002Figma plugins + Claude Code restructuring files, and the plugin i wrote
003Figma Make quick directions to react to

← previousManagerbot next →Record Pull