AI Agents Just Unlocked $7 Trillion in Invoices Nobody Could Finance

Trust is the most expensive thing in business, and nobody bills you for it until it is gone. Four nine-figure receivables frauds hit the industry in a single year, and every one of them traces back to the same quiet failure: a lender who built a relationship, got comfortable, and stopped checking. Anthony Eden watched that pattern from inside one of the world's most secretive private equity shops, and he decided the fix was not more clerks making phone calls. It was agents making thousands of them.

Funding stage: Pre-Seed. Anthony states on air that Iridium raised its pre-seed in February following an accelerator, with revenue now contracted but no seed round announced, so Pre-Seed is the accurate public call even though he hints at news coming.

Anthony is the founder of Iridium (iridiumcredit.com), a company that automates the entire lifecycle of invoice finance for lenders: verification, fraud detection, collections, and cash reconciliation. In most of the world, borrowing against invoices is the default way businesses get short-term cash. In Latin America and parts of Europe it runs around 15% of GDP by Anthony's telling, with Belgium as high as 22%. In the US it is roughly 2%, because without government e-invoicing registries, verifying that an invoice is real means emails, phone calls, and logging into accounts payable portals by hand, at a cost Anthony puts at about $22 per invoice. Iridium replaces that grind with email agents, voice agents, and browser agents, so lenders can profitably finance invoices they used to turn away, and the 58% of SMBs who Anthony says get denied the credit they seek finally get a shot.

The origin story is just as good as the product. Anthony started as an intern at Cerberus Capital Management, came back to build credit underwriting automation alongside PhD scientists who wrote his college textbooks, watched it print millions for European banks, and realized someone was going to compress financial services into a hyper-efficient market. He decided that someone would be him. He joined an accelerator a month late, teamed up with co-founder Preesha Gehlot, an ex-Bloomberg and Microsoft machine learning engineer who published two AI papers before graduating from Imperial College London, and closed their first customer in a two-week sales cycle. Today the four-person, fully Claude Code pilled team is packing for New York with a contracted book of revenue and a pipeline Anthony describes as massive.