Day 2: One of our most common requests from customers was to automate internal tools and cloud software that didn't have APIs or MCPs. Last week we launched that feature in @grepdotai. Today we're open-sourcing the harness feature that makes it work as an @pidotdev extension.

We're building self-improving agents to automate tedious, repetitive operational work @GrepdotAI. At IG, a UK online brokerage, Grep agents review compliance alerts, helping over 80% of traders onboard without manual intervention. IG is now a FTSE 100 company. Operational work often means following a specific procedure, gathering information from different sources, checking it against what's needed, and deciding what to do next. The steps are repetitive, but no job is the same, making the work tedious to do by hand and hard to automate with workflows alone. Meanwhile, using frontier agents to reason through every task is often too expensive to scale. Grep.Al gives you an agent workforce that: -> Takes minutes to create and are domain experts with the right skills and tools from day one. -> Learn from every run, moving repeatable steps into code and workflows and saving frontier intelligence for exceptions and unfamiliar cases. -> Get smarter and cheaper as they scale using our AgentRun harness. -> Runs on our managed agent platform with secure sandboxes, audit logs, evals, cost controls, team sharing, and access controls included. Grep AI agents are available on demand and can run on schedules or in batch. Access them via the console, MCP, API, or SDKs that integrate into your existing workflows. Our customers, including IG, Airwallex, Worldpay, FV Bank, and Citizens Bank, have saved more than 250,000 hours of operational work with Grep Al agents across onboarding, doc verification, compliance, underwriting, operations, HR, GT ops, market research, and business intelligence. If operational bottlenecks are slowing your business down, create an agent workforce with Grep to automate them in minutes. P.S. We're in the @ycombinator Fall 2026 batch!

A customer asked me last month how other companies are rolling out AI agents. What's working, what it costs them, and where things stall after the pilot. I answer some version of that question on calls most weeks. I've never written any of it down to share with others, so I'm restarting my Substack newsletter. I started the original newsletter, The Hitchhiker's Guide to AI, a few weeks after ChatGPT launched, when everyone was still working out what generative AI was. It let me learn about the space as an outsider looking in. Writing about AI every week eventually led me to the conviction to found an AI startup, and the writing took a back seat. That was three years ago. The newsletter is now back as The Operator's Guide to AI because I'm no longer an outsider looking in, and neither are you. We're all insiders, trying to build, deploy, and scale AI systems, and I want to share as much as I can about what we're learning along the way @grepdotai. But there are a million different AI publications, so why should you subscribe to mine? 1/ We've been deploying agents into production since 2023, worked with every generation of models, and made plenty of mistakes along the way. I'll be sharing a lot from first-hand experience, not only of deploying agents at large enterprises, but also of how we use AI internally too. 2/ We speak with large enterprises all the time about their AI transformation efforts: no one has really worked it out, and everyone is a lot earlier than you think. I'll be sharing what we're seeing on the ground. 3/ There's a lot of noise in the AI space, with new models launching every week, new capabilities being announced, and endless fundraising announcements. I want to cut through the noise and focus on what practically matters for folks deploying AI and agents at scale. If that sounds interesting, you can subscribe below... The first post on why agents fail in production goes out this week.

We’ve been busy!

At @ParchaHQ we saw coding costs explode so we built our own IDE that does routing at the harness level e.g. Claude Code for planning, Codex for long running execution, Cursor for bugs. Plus custom routing, remote sessions, a browser inspector and voice interface. Now it's open-source! 👇