Hatchet

If you've signed up for our cloud platform, then you've received this welcome email from our CTO, @gabe_ruttner! 💙 Our latest cookbook explains how we set up this durable workflow with @loops production email delivery. P.S. Gabe really does read every single reply.

Link here 👉 docs.hatchet.run/cookbooks/pdf-…

PDF pipelines are a natural fit for DAG workflows. Our latest cookbook with @reductoai shows you how to build one that's production-ready. 🧑‍🍳🍳

New @hatchet_dev HQ in Brooklyn! We started with a 3-person space in January and outgrew it quickly. Remote-first Hatchlings from across the EU and US (Boston, SF, Philadelphia, Amsterdam, and Luxembourg) are now regularly working in-person with us here in NYC.

JD here 👉 ycombinator.com/companies/hatc…

🚨 We're hiring a Growth Engineer! 🚨 Here are some thoughts from a totally unbiased focus group on why you should join our team.

we're hiring a growth engineer at @hatchet_dev! if you're a talented, ambitious person looking to work in startup growth alongside truly world-class product engineering talent (who also happen to be brilliant, kind, and hilarious), here are a couple things you should know... 🧵

The team @getenter_ai builds AI agents that process hundreds of thousands of legal documents for major financial institutions and corporations. Last week, they became Latin America's first AI unicorn. Here's how Hatchet helps power the orchestration layer underneath ↓

I am a hypocrite. Feels good to get that off my chest. A few weeks ago we shipped OTel and Logs so @hatchet_dev users could reason about their system state. Meanwhile, we needed a better pulse on our own. Here's how we rebuilt collecting our logs, spans, traces + metrics... 🧵

If you've signed up for our cloud platform, then you've received this welcome email from our CTO, @gabe_ruttner! 💙 Our latest cookbook explains how we set up this durable workflow with @loops production email delivery. P.S. Gabe really does read every single reply.

Had an interesting conversation with a user recently on how we built retries for durable tasks in @hatchet_dev. A couple thoughts... 🧵

@abelanger5 🎨

@hatchet_dev What is this

😤

@abelanger5 And for the parts that do parallelize cleanly, there's Hatchet → hatchet.run/use-cases/para…

Running 10 coding agent sessions doesn't make you 10x faster as a developer. Here's why, via Amdahl's Law (via @abelanger5) ↓

Welcome to the team, @igor_kupczynski! 💙

April was an unbelievably great month at @hatchet_dev: ➡️ We ran just over 3 billion tasks, a 27% increase over March. By comparison, when we first launched Hatchet Cloud two years ago, we ran 10k tasks the first week of launch. Now we regularly execute that number of tasks per second. ➡️ We increased user signups by over 25% compared to last month. ➡️ Downloads of our Typescript and Python SDKs have more than doubled in the last 2 months. ➡️ We've started regularly posting cookbooks on our docs site, including guides on how to build a support agent and a welcome email using Hatchet. ➡️ Coming off our March launch week, we ran our first "cooldown" month, getting a number of bugs and optimizations shipped, along with two major internal projects: a centralized control plane and federated observability into all 4 of our regions. Below is an outtake of a team picture we took back in February. It's hard to believe that was taken just a few weeks ago given how quickly things are moving. Excited for May. Some big things coming soon. 🚀

We're finally releasing a public roadmap 2.5 years into building @hatchet_dev. But why did it take so long? A couple thoughts. In the early days, decisions and priorities on what to build next would stem from small signals from early customers. This pulled us in a lot of different directions and maintaining a roadmap when things would change so frequently was nearly impossible. Also, making those commitments when resources were tight felt irresponsible. I'm reminded of a talk @mwseibel gave in the @ycombinator S20 batch about Twitch and how they didn't maintain a roadmap. Instead, they'd build things that kept re-emerging. I think this operating model works well for early-stage startups... until you have larger customers with longer decision-making cycles. Those larger customers are often evaluating what and where, exactly, you are today and importantly, whether you'll be able to provide what they need in the future. Anyway, we're mature AF now. 😎 Check out what we're building this quarter and let us know what else you'd like to see!

Full update here 👉 linkedin.com/pulse/whats-ne…

We just shipped a complete rewrite of our durable execution layer, including: 👉Pause + resume execution 👉Full OTel observability 👉Child task spawning for subagents and tool calls 👉Support agent cookbook + Our Q2 Product Roadmap is LIVE, featuring some new experiments! 🚀🧪