Zack Proser

Climate change is a crisis — but we have solutions to start addressing it. I’m now funding a project on Wren to offset my own CO₂ — if you join me today, Wren will protect an extra 5 acres of rainforest! wren.co/join/ZackProse…

One of the secret best parts of MCP Night is the underground chess tournament during the after party ♟️ Not to be missed! 😊 Starts in 1 hour at The Regency Ballroom. See you there!

MCP Night Agent Mode starts soon…

I'm a 90 WPM typist. With WisprFlow I hit 184 WPM - more than double 🚀 Works in Cursor, Claude, Slack, Twitter, everywhere. First voice tool that actually works for developers. zackproser.com/blog/wisprflow…

@rodrigoelias 🙇 Means a lot. Thank you

@zackproser This blogpost inadvertently made me aware of this essay of yours and I can’t explain how much every word reasoned with me. Very inspiring. TY zackproser.com/blog/executive…

Moved my AI assistant from a box under my desk to always-on AWS infra in a day. EC2 is cattle (replaceable in 2 min). EBS volume is the pet (memory, skills, personality persist across rebuilds). OpenTofu, Tailscale, zero public ports. zackproser.com/blog/building-…

Got a pitch that opened with Hi {{first_name}}, — the template variable literally unrendered. So I built a thing that doesn't bother either: a brutal AI screener. Three examples in post. "We don't template people we really care about, right?" zackproser.com/blog/inbox-cla…

I consistently hit 179–184 WPM dictating code with WisprFlow — more than double my 90 WPM typing speed. 182,718 words across 36 apps later, it's become essential. Here's my full review: zackproser.com/blog/wisprflow…

For the past several months, our engineering team has been building a new agentic coding system designed around a simple idea: the self-driving codebase. These aren't just agents to write code on demand. They detect triggers, spin up secure sandboxes, gather dynamic context, open PRs, verify their work, unblock the next task, and learn from every failure. We think the future of software engineering may look more like swarms of event-driven agents, running continuously, with humans setting direction and reviewing the work. Ramp built Inspect. Stripe built Minions. Spotify built Honk. We call ours Horizon. 🌅 Read more about what we learned and where it's going next:

Most AI agents run with borrowed sessions and way more access than they need. Your support agent looks up one invoice using the rep's full OAuth session, the one that can also refund, delete, and modify. Agents need their own identity. Full guide: workos.com/blog/ai-agent-…

@hadi_hlk @nicknisi YOU ARE WELCOME SIR - DONT USE PERSONAS THEYRE WEIRD. DONT USE ANY ENVIRONMENT EXCEPT TAHOMA LIGHT OR MAYBE DARK IF ITS LATE LOGGING INTO YOUR STREAMING APPS SUCKS YOU HAVE TO GET A THIRD PARTY CLIENT FOR SPOTIFY STILL WORTH IT

I blame @zackproser and @nicknisi for this purchase 💸

workflows running on workers help workers at WorkOS do work

Our blog bot wrote this post. It also pitched the topic itself. @Cloudflare Durable Workflows turned what would've been queues + workers + callbacks + checkpointing into step functions. A multi-model AI pipeline runs durably behind every Slack message, and our team is now publishing like crazy - getting engineering stories and announcements out in minutes that previously would have sat in the backlog for weeks. workos.com/blog/cloudflar…

WorkOS is now a supported provider in @stripe Projects. One command from the terminal provisions a real WorkOS environment, writes your credentials to .env, and loads agent skills so your coding assistant already knows how to scaffold the integration. Learn more: workos.com/blog/workos-jo… Try it on Stripe Projects: projects.dev

Colab Pro wasn't enough for me to fine-tune Llama 3.1. Ran into constant OOM errors and had to buy extra credit packs. Lightning.AI Studio was the only option that just worked. I compared all the major cloud GPU services for Jupyter notebooks: zackproser.com/blog/cloud-gpu…

OpenAI Codex went from 40-60% success rate to 85-90% on well-scoped maintenance tasks. That's not a hype number — that's the difference between using it occasionally and building your morning routine around it.

@nicknisi @ClaudeDevs Get em big dog

Hey @ClaudeDevs Auto mode is unusable if it's going to decide not to call a tool and just hang without actually telling me that.

@nicknisi @ClaudeDevs Yeah!!!

@ScalingDevTools @nicknisi @WorkOS Thanks @jacksbridger ! Was great to see you

When @nicknisi and @zackproser from @WorkOS ran their workshop at AIE, someone just got their Hermes bot to complete it and submit - and it was good

In case you’re wondering why I haven’t shipped uilogue and my other side projects yet, exhibit A: Me spending hours perfecting the agent cursor asking you for a high five after you join the waitlist. ¯\(ツ)/¯

@WolframRvnwlf @ToolUsePodcast @MikeBirdTech @karan4d @NousResearch Awesome. Really enjoying hermes so far!

New @ToolUsePodcast episode by @MikeBirdTech is live and I had the pleasure of being part of it alongside @karan4d from @NousResearch and three other great builders. We went deep on Hermes Agent: self-improving skills, ambient agents, why the harness matters as much as the model, and why open source AI must win. I got to share some @WolfBenchAI insights and real-world use cases. Give it a listen on Spotify or watch it on YouTube - links in reply.

ACQUIRED Unplugged with @bcherny and @GarrettLord 🚀