Grep AI joins YC F26 to help teams improve repetitive work
Grep’s agents turn repeatable steps into reusable workflows so teams can scale work that still needs judgment.
Grep is officially launching as a YC company in the Fall 2026 batch. At Grep we're building self-improving agents for repetitive work that requires judgment.
At IG.com, a global online brokerage, Grep agents helped increase automated onboarding to over 80%, leading to 2× more first trades their entry into the FTSE 100.
Built for the work that scales your business
Across underwriting, due diligence, document review, and operations, teams repeatedly gather information from different systems, check it against requirements, and decide what to do next. The steps repeat, but each case can bring exceptions.
Rigid rules can break when an exception appears. Agents that reason through every familiar step from scratch can be expensive to run at scale.
Agents that improve with every run
Grep agents learn from runs of the same task. As they identify repeatable steps, Grep moves more of that work from model calls into code and workflows, leaving the agent to focus its reasoning on exceptions.
Teams can run and tune the process across more cases. The goal is for an agent to become more useful and more economical as it gains experience with a workflow.
Our customers
Customers also include Airwallex, Worldpay, Shopmonkey, FV Bank, and Citizens Bank. Their teams use Grep for work across underwriting, due diligence, document review, and operations.
Why we’re building Grep
Founders AJ Asver and Miguel Rios-Berrios previously built AI products and infrastructure at Coinbase, Brex, Google, and Twitter. At Brex, they led a team that automated risk, compliance, operations, and customer support before AI agents existed.
They started Grep to help rid businesses of tedious, repetitive work that slows teams down and leaves people less time for work that needs their judgment.
Create your first agent at grep.ai, and we’ll help you tune it for your workflow.