Building @GrepdotAI (YC F26). Prev Product @compound, @brexhq, @coinbase, @google. 2x @ycombinator founder, @uniofoxford. Dad of twins, in-person maximalist.
San Francisco
Joined January 2008
- Tweets3.3K
- Following1.4K
- Followers4.1K
- Likes2.5K
AJ Asver retweeted
Day 3 of open sourcing our harness as pi-extensions: pi-durable-disk and pi-transport. We use both to run agents from anywhere. For us, the agent is the workspace. And the workspace lives in @archildata.
With @pidotdev durable, the session DB also lives in the disk. We can mount and unmount it from its sandbox, and run the pi loop from a web browser (w WASM) or a @modal VM with GPUs - sometimes in the same session.
Even easier, with Archil's serverless execution, we can run simpler agents. For example, we have an agent in-process go through 100s of GBs in Claude Code sessions and recall the entire history of a project in seconds.
AJ Asver retweeted
Honestly I don’t want to check out any part of a codebase. I want to just point my agents at a codebase, tell them to make changes, and let agentic checks and evals pass. If I’m running my agents in the cloud, and the code is hosted in the cloud, why check out anything at all?
AJ Asver retweeted
Proud to power such an awesome product!
Day 3 of our week of launches: Connected accounts + authenticated browser access.
Grep already includes over 100 data integrations out of the box - no logins needed. But as our agents took on more automation, it became clear that customers needed a way to add their own integrations, APIs, and even internal tools with no MCP or API connection.
That's why we're excited to team up with @composio to bring over 1,000 integrations to @GREPdotAI for all your favorite SaaS products, including Gmail, Calendar, Outlook, Notion, Salesforce, and much more. Customers can also add their own custom APIs and MCPs to Grep's agents for tools without a built-in integration.
But not every product or application has an API or MCP, like internal tools, legacy software, and government-run databases. For those use cases, we're also launching authenticated browser access for agents.
Just log in to any web app once, and we securely store your credentials in an encrypted vault so your agents can use the application for automation. If you don't want your credentials stored, you can also log in manually every time the session expires.
Now, Grep agents can automate any online application or website and access any online data source, reducing the repetitive work that slows down your business.
For example, a customer onboarding onto a fintech might hit a manual review step because a compliance screening alert flags their name as similar to one on a sanctions list. An agent can pull compliance alerts from a case queue in Salesforce, log in to a legacy screening solution like Nice Actimize to disposition the alerts using deep research and Grep's data integrations, and then send RFIs to customers via email when there isn't sufficient evidence.
Check it out at Grep.ai
AJ Asver retweeted
Day 4: What's an agent without evals?
Today we're launching native evals in @GREPdotAI. Now every time you create an agent, we automatically create a set of evaluation questions that you can edit and update. Then, while the agent is running, your team can evaluate it and provide feedback on its accuracy.
That feedback goes back into the agent harness, and Grep suggests prompt updates you can apply like a GitHub pull request to improve your agent.
AJ Asver retweeted
Our coding agents mention each other in Slack without being prompted. They ask for support / context (each agent has access to their owners' sessions and the company brain). It's a bit eerie, tbh.
AJ Asver retweeted
Today we're launching artificial intuition.
A model that predicts which idea will stand out, before anything is published.
Already live in Native for 2,200+ businesses.
API waitlist is open: native.no/norn 🚀
AJ Asver retweeted
Replacing LLM calls with deterministic code is a no brainer and also results in better code that better matches what users want
We built a new harness using @typesafeai's Jev that cuts the cost of repetitive work by 90%. The harness learns the job as it runs, moving steps from LLM calls to code.
Running 100,000 compliance alerts costs >$290K on Opus 5.
With agentrun() we got it down to <$26K.
AJ Asver retweeted
Love this. One of the tools I built for our team @grepdotai is an artifact server. It has similar concepts, except it can also publish back to the Claude Code sessions through @herdrdev. Here's a short demo of how we use it.
AJ Asver retweeted
my dad and i built a robot for your parents.
today we’re launching @rhem_labs
our family spent nearly 20 years caring for my grandmother and aunt at home. we saw small health changes go unnoticed between check-ins. and simple things, like making a call or booking an appointment, meant waiting for someone else to help.
we built the robot we wish they’d had.
your parent just talks to Rhem. it helps make calls, book appointments, manage reminders, and get everyday things done online.
through the family app, you can follow changes in health readings and routines, see what Rhem handled, and step in when you’re needed. your parent chooses what to share.
Preorder Rhem Today → rhem.ai
AJ Asver retweeted
Today we're reintroducing @quivlyai, backed by @ycombinator.
Quivly is the AI workforce for everything after the sale, built especially for companies with usage-based pricing.
Teams using Quivly take customers live in 7 days instead of 30, then expand them 4.3x on average.
Our agents turn kickoff calls into launch plans, chase blockers, catch risk and expansion signals, and have the full account story ready at renewal.
Closing the deal is just the beginning. Quivly handles everything after that.
AJ Asver retweeted
Harnesses from labs have an incentive to burn tokens, which means harnesses from startups have a real utility, like how Grep can automatically swap token burn into deterministic tested code that is repeatable just by observing agent use
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!
AJ Asver retweeted
Highest leverage thing I've done as a technical founder: getting my agents to run without my attention as much as possible (see previous post about this in the comments).
The obvious instinct now is to run more agents. Instead, I'm spending that time and attention on GTM and being in front of customers.
My job as a founder isn't to maximize lines of code or PRs landed. It's to put my attention wherever the company needs it most. With frontier LLMs removing so many of the bottlenecks on the technical side, that matters more than ever.
Writing this from the HQ of one of our bank customers in Puerto Rico.
Early observation: Opus 5.5 is completely changing the way I work.
1. For large tasks, I'm spending more time with Fable upfront: understanding the problem, reading discovery docs, TDDs, implementation plans, etc. For smaller ones, I just use Opus 5.5 to do the next steps end to end.
2. I give Fable a version of this prompt: "You are the tech lead manager. You have a team of very capable staff and senior engineers. You will give each of them a clear subsystem of this task. They will write code, review it, run e2e tests, iterate, and deliver it to you. You then validate it or send it back with feedback until it's done. When all the work is done, you put the system together and run e2e tests. Go back to the engineers if needed."
3. Fable spawns many Opus 5.5 subagents to do the work.
4. I end up with a stack of PRs to review. We have a "risk" system with rules that either auto-merge after automated checks pass (including autoqa) or require human review. At this point, most PRs auto-merge.
5. The result: almost twice as many PRs merged with no noticeable issues, from the same number of prompts. Since PR size has stayed roughly the same, I read this as getting far more done per prompt. The obvious trade-off is that I'm reading less code. I'm OK with that.
Big props to @chrisrodz35 who put our factory to work so we can do this responsibly.
AJ Asver retweeted
My @muse was taking too long on finding the delivery fee of a fridge. I checked the browser to see what was going on…
THE AUDACITY!
AJ Asver retweeted
Today, I'm excited to bring @OasiveAI out of stealth and announce that we're backed by @ycombinator!
Oasive helps bond investors find better bonds to buy.
Uncover opportunities you might otherwise miss, compare investments and test the risks before putting on a trade.
Which Treasury maturities look cheap? Is there a curve trade worth putting on? Macro and Rates Research helps you explore both. In mortgage-backed securities (MBS), compare pools and see how changes in rates and prepayments affect your portfolio.
Our AI analysts bring the research and proprietary models together, so you can go from idea to decision in minutes.
Thank you to @kul and @garrytan for taking a chance on me and believing in @OasiveAI!
Start your Macro and Rates Research free trial or request an MBS demo: oasive.ai/join
Here's what you can do 👇
AJ Asver retweeted
The ceiling on what we can do is so much higher now!
I've intellectually known that software engineering was going to go away but it's just started to feel real.
The little dopamine rushes — learning a language, reconfiguring keybindings, upgrading packages or pushing around code on a page to make it easier to read — fade in the rear view mirror, vestiges of a profession no longer needed.
The job is more director than architect or builder, and quickly rising to higher levels of abstraction. Agents may do the work of spewing code into files and servers, but building ambitious real world software is hard in a new way. We are responsible for houses we did not create and whose floors we've never truly walked.
The longer you've been an engineer in the previous world, the harder the transition is now. But it's time to rewire your brain to appreciate the new ways of how things are going to be built.
The ceiling on what we can do is so much higher now.
AJ Asver retweeted
Almost every Jev usecase coming up feels far far more creative and useful than 90% of things people build with LLMs.
what if copy/paste was smart?
powered by @typesafeai jev
it feels like every computer interaction will get rewritten
