Sticker

Benchmarks and RL Environments for Procurement

[Sticker @StickerCommerce]

https://t.co/jalKJBfZCs

[Sticker @StickerCommerce]

procurement n. /prəˈkyo͞orm(ə)nt/
How a business finds, negotiates, buys, and pays for the goods and services it needs.

Today we’re launching Sticker, a research lab building environments and benchmarks to train and evaluate AI agents in procurement.

We think procurement is one of the most valuable things AI will learn. In the US, more than $20 trillion a year goes to buying goods and services from other businesses, and much of it moves through a slow, manual process of finding suppliers, comparing quotes, negotiating terms, getting approvals, and reconciling invoices. It’s long, multi-step, high-stakes work, and frontier models still can’t do it reliably.

Before building environments, we built and ran procurement infrastructure for real businesses, so we know where this work gets hard. Our research simulate procurement from request to payment, on the same portals and software buyers use, and agents inside them must handle backorders, follow purchasing policies, and resolve invoices that don’t match what was ordered or received.

When AI can do procurement reliably, businesses will buy faster and waste less, suppliers will get paid on time, and the people doing this work today can focus on decisions instead of paperwork. We’re building Sticker to make that future arrive sooner.

Yahya Rao, Suyash Unnithan
October 15, 2026

Sticker is backed by Better Tomorrow Ventures, Slow Ventures, and angels.