About GREP AI

More time for
what matters to you.

We’re building Grep to take repetitive work off your plate, so you have more time for the people you help, the ideas you want to pursue, and life beyond work.

Our story Read our lessons

Our story

A small team. A bigger purpose.

We started with a problem we knew firsthand: too much of the working day gets lost to repetitive tasks. Here’s how that became our work.

2023
We started Parcha.
At Brex, AJ and Miguel saw how much time teams spent on manual operations. We founded Parcha Labs to help, working with our first customers on compliance and due diligence: real work, with people depending on the outcome.
November 2023
We started with one specific use case.
We focused on business verification: helping teams understand who they were doing business with. Starting with one specific use case let us learn the work deeply and build agents that could handle it reliably.
December 2025
We opened the door to more teams.
We opened Grep as a research preview, bringing what we had learned to work beyond compliance. Within two weeks, hundreds of people across consulting, technology, and financial services were using it. The need reached far beyond where we started.
Today
The ambition is still personal.
We’re building toward a day when repetitive work takes less of yours. More room to help a customer, follow an idea, or close your laptop on time. That’s what we want Grep to make possible.

What the experience taught us.

These are the decisions we return to as we build.

01
Judge the system over repeated runs
One good answer tells you little about the next hundred cases. We care about consistency, recovery, and evaluation across the work an agent actually encounters.
02
Give agents a proper workspace
Instructions, files, tools, and accumulated context help an agent complete substantial work. Its environment deserves as much attention as its prompt.
03
Make the work inspectable
People need to see the evidence and decisions behind an outcome. Clear traces help teams review results, understand mistakes, and improve the process.
04
Make the economics work at volume
An agent becomes useful when you can afford to keep running it. Reuse successful procedures, move repeated steps into code, and spend on reasoning where it contributes.

Built by Parcha Labs.

We bring product and engineering experience from Brex, Google, Coinbase, and Twitter. Since 2023, we’ve been applying it to the practical problems of putting agents to work.

AJ Asver

Co-founder & CEO. AJ built AI products at Google, led Data Product at Coinbase, and was Director of Product at Brex. Seeing the operational bottlenecks inside growing businesses led him to start Parcha and work on agents that could take on those processes.

Miguel Ríos

Co-founder & CTO. Miguel led Global Data Science at Twitter and held leadership roles in data, platform engineering, and AI strategy at Brex. At Parcha, he brings that experience to the engineering work behind dependable agents, and shares what the team learns along the way.

From our notebooks