Injester

Injester runs AI shopping agents through an online store's catalogue the way a human shopper would, recording where products fail to be found, compared or bought, then tracing each failure back to the underlying product data so merchants can fix it and retest. A Commerce API exposes the store to agents over MCP and REST.

As consumers increasingly delegate product discovery and purchasing to AI assistants, a catalogue that reads well to humans can be invisible or broken to agents. Injester opened its beta in October 2026, betting that agent-legibility becomes a line item merchants actually test for.

The company says merchants need their catalogues to work for agents as well as human shoppers — a gap that only shows up when you make an agent try to shop. Injester's answer is to surface those failures and tie them back to fixable product data.

Co-founder Benjamin Shyong was a staff manufacturing technical program manager on Meta's AR/VR hardware (Ray-Ban Meta) and earlier worked on iPad Pro at Apple, with an industrial engineering degree from Purdue; he's building Injester with co-founder Vishal Verma. It's a hardware-and-process background pointed at a software-and-data problem.