Every chapter has an end.
This one is a happy one.
77Labs began as an idea on a whim and became something much bigger: a small but mighty fund and holding company built on ownership, experimentation, and operational conviction. Over eight years we owned and operated 11 businesses and turned around 70% of them. We also took one quantitative investment strategy from research to market. That's a record we're deeply proud of, and it happened alongside the investors, partners, operators, and founders who backed the hard decisions.
Those years taught us the same lesson again and again: the value was never in the software alone. It was in the problem being solved, the operations behind it, and staying close to the people who actually use what you build.
That truth has only sharpened. AI is pushing the cost of building software toward zero, and it's reshaping the moats that defined the last decade of SaaS. Features are becoming commodities, and markets that once took years to consolidate now shift in months. We'd rather move with that current than hold onto a version of the market that's already changing.
So we exited deliberately, on our terms. We handed the strongest businesses into their next chapters, wound down the ones that didn't work, and are closing ours with the same spirit that built 77Labs in the first place: curiosity, conviction, and a belief that the best businesses are built closest to the real world.
By the end of 2025 we'd narrowed the portfolio to five. Today, every one of them has been exited or shut down, and with that, 77Labs' operating chapter is complete.
Eight years.
One hundred and thirty-six repositories.
From image-classification microservices to AI-native desktop apps — a record of what the team built and shipped over eight years.
Where we started. Where we ended up.
The first commits and the last — same team, same instincts, vastly different surface area.
Computer vision, scrapers, classification services
Python microservices that looked at images and decided what they were. They became the computer vision behind our resale and pricing technology.
- Bag Classification Service
- Color Detection Service
- Brand Classification Service
- Model Classification Service
- Scrapers— Set of scrapers for specific sites
- Markup Tool Backend
- Retail Dashboard Backend — maintained through 2026
AI-native products, end-to-end
Full-stack TypeScript, Python and Rust: research platforms, desktop apps, APIs and dashboards, owned end to end.
- Alternative-data research — quantitative investing
- AI desktop apps — Tauri + Rust
- Healthcare and patient apps
- Commerce dashboards and marketplace tools
- AI workflow automation
Every repository, plotted in time.
Each dot is a repo, positioned by when it was created and colored by primary language. Open dots are archived chapters; filled ones were still in use in 2026.
Swipe sideways to see every year
The stack told its own story.
Python carried the early CV work. JavaScript and Kotlin powered the storefront era. TypeScript took over once we owned product. Rust showed up the moment we shipped desktop software.
Three eras, three stacks
Python dominated 2018–2019 — classification services, scrapers, data tooling.
JavaScript & Kotlin took 2020–2021 when we shipped storefronts and mobile apps.
TypeScript became the default from 2022 onward — about 70% of everything built after that.
Rust arrived in 2026, when we finally needed it.
The curve never stopped going up.
Total repositories created, year over year. The team built at its fastest pace in its final year.
Four repositories from the earliest era stayed in use to the very end.
Most teams' early work gets archived. Ours got compounded: these systems ran underneath business after business, crossing 5+ years of continuous evolution — the longest, Retail Dashboard Backend, received commits for 8 years straight.
From scraping pages to building autonomous systems.
Same team. Same instincts. Vastly bigger surface area. Thank you to everyone who built it with us.