Signups
1
MRR
£0
Stack
Flutter desktop · Firebase · Direct .dmg (signed + notarized)
- Current read
Faruq + Amina · Aug 29, 2026
Faruq thesis
Receipts is a Mac-native capture-to-content tool that refuses to become another generic productivity app by focusing on the unique needs of builders who ship code.
Next moves
- Integrate with popular code editors and version control systems to streamline capturing work in progress. - This unblocks the thesis by making it easier for developers to capture their work as they build, without interrupting their workflow.
- Develop AI-driven content suggestions to help users turn raw clips into polished posts. - This addresses the core pain point of editing rabbit holes and helps users share their work more efficiently.
Watch
- Over-reliance on AI for content generation could alienate users who prefer manual control over their posts. - If users feel that the AI is taking too much control, they may abandon Receipts in favor of tools where they have more creative freedom.
Amina visual track · running
Use real product captures first; create hero stills, proof crops, and social/OG variants from Faruq's latest thesis and current-state evidence.
— Why it exists
I built Receipts because I couldn't keep posting consistently — and I'm not the only one.
The problem isn't writing. The problem is recall. By Sunday, the most postable moments of Tuesday are gone. The bug that took two hours and taught you something. The decision at 3pm that changed product direction. The frustration that turned into a breakthrough. All of it forgotten by the time I sat down to write.
Existing tools solve adjacent problems. Note-taking apps are for reference. Content schedulers start from 'you have something to post.' Journaling is too heavy for in-the-moment capture. Nobody was doing the loop builders actually need: capture in the moment → surface what's worth posting → ship.
So I built it. Mac-native first because nudges need reliable system notifications — web fails 30-40% of the time and that kills the product.
— How it's built
Cut the free tier in favour of card-on-file 7-day trial
Free tiers create permanent users carrying Firebase + AI costs without revenue. 7-day trial with card-on-file converts at 60-80% vs 1-5% for free-to-paid. Bet on revenue proof over user-count proof for Week 3.
Mac-native first over web
Web notifications fail 30-40% of the time. The nudge mechanic IS the product — unreliable delivery would kill it. Accepted Mac-only audience cap to ship a working product instead of a broken cross-platform one.
Provider-agnostic AI from v1
Built the abstraction now even though it added time, because cost optimisation pressure (Deepseek) and quality optimisation pressure (Claude for angles) are both real. Cheaper to architect right than to refactor in v1.1.
Captured user's own voice over generated content
Post angles draw hooks from user's actual captures, not AI-generated copy. The whole product fails if the angles read like ChatGPT-style content. Locked the angles prompt Friday night with real-data testing before letting it ship.
— The lesson
“Card-on-file at trial start was the right call architecturally. Trial signups will be smaller than a free tier would have produced, but the conversion math is decisive — fewer users, more revenue, less infrastructure burn.”