The data layer for the recycling trade - from a first-of-its-kind recyclables marketplace to AI that reads material quality from a single photo.
SAFI, LONDON. Founded 2021 as TrueCircle. A recyclables marketplace that grew into an AI quality-control company - built by operators from Revolut, Onfido and Monzo who decided the circular economy needed better data, not better intentions.
Recycling has never suffered from a shortage of good intentions. It has suffered from a shortage of good data. When a recycler cannot prove what is inside a bale - how much of it is clean PET, how much is contamination - they cannot price it, plan around it, or fully trust it. That uncertainty ripples through every trade, and it is a large reason roughly 90% of plastic never gets recycled. Safi exists to close that gap.
Safi is a London-founded climate-tech company that builds software for the recycling industry. It launched in 2021 under the name TrueCircle and rebranded to Safi in 2023. Its first product was the recycling sector's first digital B2B marketplace: a place where sellers of recyclable plastics, paper and metals could be matched with buyers across borders, with quality checks, logistics and financing built in.
That marketplace grew quickly - from zero to more than $40M in annualized trading volume within roughly a year. But the deeper the team went, the clearer the real bottleneck became: quality data. So Safi doubled down on it. Today the company is best known for AI quality control - computer vision that reads the composition of a bale from an ordinary phone photo in about three seconds, at roughly 95% accuracy, without anyone having to break the bale open.
The result is a continuous, real-time picture of material quality where the industry previously relied on sampling, estimates and gut feel. For a recycler, that means fewer surprises, better pricing, and higher recovery rates. For the planet, it means more material actually gets recycled instead of landfilled or burned.
It is unglamorous infrastructure - and that is the point. Safi is not asking anyone to care more about recycling. It is trying to make the accurate choice the profitable one.
The recyclables trade moves hundreds of billions of dollars of material a year, much of it still coordinated by phone, email and spreadsheet. Opacity at each step quietly erodes value.
Figures are company-stated and approximate. Bar widths are illustrative.
Analyze material composition from a phone photo in about three seconds - purpose-built for PET, HDPE and used beverage cans, no bale-breaking required.
Check composition and contamination across incoming and outgoing loads so quality issues can be caught and acted on immediately.
Use composition data to guide bale selection and predict yield before a production run begins.
Benchmark suppliers on the quality they actually deliver, turning buying into a data-driven decision.
The industry's first digital B2B marketplace matching international buyers and sellers with quality checks, logistics and deal management built in.
Early payouts for sellers and extended terms for buyers, delivered through a partnership with embedded-finance provider finmid.
Material recovery facilities, recyclers, traders and manufacturers who buy and process recyclable plastics, paper and metals. Safi cites use across 80+ sites in 25+ countries - named clients include BMP Recycling, N+P Group and Ecoiberia.
Early on, Safi earned a commission on trades plus margin on embedded financing. Its current focus is a SaaS model - selling AI quality-control software to recyclers and procurement teams on subscription.
Rather than expensive fixed sorting robots, Safi reads quality from a phone photo - lightweight, deployable across many sites, and focused on the trade decisions that hardware alone doesn't touch.
The team blends computer-vision experience from identity unicorn Onfido with operating experience from Revolut, Monzo and Unilever - a rare mix pointed at an old-line industry.
Where it fits. In recyclables marketplaces, Safi's peers include Cirplus, Recykal, ScrapMonster and Circular. In AI vision and sorting, the competitive set includes Greyparrot, Recycleye, AMP Robotics, TOMRA Systems and Pellenc ST. Safi's position sits between the two: less a sorting-line hardware vendor, more the quality-and-decision layer that helps recyclers know exactly what they have and trade it accordingly.
The company is founded in London to bring data and transparency to recycling.
Raises $5.5M led by Lowercarbon Capital and launches its recyclables B2B marketplace.
TrueCircle becomes Safi, raises a Series A led by Nosara Capital, enters the US, and partners with finmid on embedded finance.
Publishes customer case studies and expands its computer-vision product for PET, HDPE and cans.
Sharpens its identity around AI quality control, cited across 80+ recycling sites in 25+ countries.
| Round | Date | Amount | Notable investors |
|---|---|---|---|
| Pre-seed | Mar 2022 | $5.5M | Lowercarbon Capital, Passion Capital, Giant Ventures, firstminute Capital |
| Series A | Nov 2023 | $19.5M | Nosara Capital (lead), Lowercarbon Capital, Transition, angels incl. Vlad Yatsenko & Tom Blomfield |
Founders. Safi was co-founded by Rishi Stocker (CEO), an early Revolut employee who moved from fintech into climate; Eamon Jubbawy, who previously helped build the computer-vision team at identity unicorn Onfido; and Chris Charles on operations. The team numbers roughly 90 people.
No official Safi-produced video channel was confirmed at publication. Links point to the company's own demo request and public search.
Safi builds software for the recycling industry. It began by operating a digital B2B marketplace for recyclables and now focuses on AI-powered quality control that reads material composition from a phone photo.
Safi was co-founded by Rishi Stocker, who is CEO, alongside Eamon Jubbawy and Chris Charles. It was originally launched as TrueCircle in 2021.
Publicly, Safi raised a $5.5M pre-seed in 2022 and a $19.5M Series A in November 2023, led by Nosara Capital with participation from Lowercarbon Capital.
Its computer-vision quality control is purpose-built for PET, HDPE and used beverage cans, reading composition in about three seconds at roughly 95% accuracy without breaking a bale.
Recyclers, material recovery facilities and manufacturers. Safi cites use across 80+ recycling sites in 25+ countries, with clients including BMP Recycling, N+P Group and Ecoiberia.
Sources: safi.co, LinkedIn, Medium, TechCrunch, Sifted, Resource Recycling, Lowercarbon Capital, finmid, Crunchbase, CB Insights, Tracxn. Figures are company-stated or from public reporting and are approximate.