People keep asking whether the agents can do the work. Mine do enough of it that the question lost its pull. The useful question is cheaper: can I spin up another me? No. That's the bottleneck.

A portrait on your homepage says "here is a guy." I wanted mine to say "here is fifteen years of work — come dig in." Those are two different bets about what a personal site is for. One sells the person. One hands over the proof and trusts it to do the selling.

My homepage used to open with a portrait of me. A face tells you nothing about the work. So I burned it down. Now the hero is my consultancy's mark — a mind on fire — rendered as living pixels, surrounded by every essay I've written as a star you can hover and read.

I run two AI agents from one phone. One handles infrastructure. One handles content. Keeping them separate isn't an accident — it's the whole reason the setup works. Separation of concerns isn't just for code. It's for agents too.

A year ago my team trialed Devin and nobody could get a single session to finish. We killed it. The verdict was unanimous: not ready. Today it's one of the most-used agents in our Slack. The gap between "impressive demo" and "reliable tool" can close fast.

You have about ten seconds to convince a user you're already solving their problem. Get the door wrong and it doesn't matter how good your model, stack, or architecture is. Nobody walks through. Time-to-value is the moat now.

Don’t sleep on this. Arjun is a brilliant engineer.

Arjun Patel@arjunkirtipatel.bsky.social · last mo.

psst! newsletter post coming out tomorrow, sign up here so it comes to your inbox! My newsletter is called Answering Machines, and I try to take a grounded approach to how LLMs work and how they will shape what we do: Sign up here: #ai #llms #reading buttondown.com/answeringma...

I once blacked out on stage in front of 4,000 engineers. Now I sleep fine before jet-lagged days of back-to-back talks and workshops in a foreign country. I want you to understand the distance, because I started about as badly as it's possible to start.

I run three AI agents that act without me in the loop. People keep asking whether agents can do the work. Mine already do. The real question is uglier: what am I willing to let one do while I'm asleep? I have a one-line answer.

I gave my website three employees. A content bot, an ops bot, and a growth bot. Each one its own repo, each deployed to Vercel on eve, all living in one Slack channel. I reviewed Vercel's eve agent framework by actually building a business team on it.

A stranger told me AI made him suspicious of all writing. He now interrogates every sentence: was this one a machine? What about this one? I understand the reflex. But that's the wrong question. Here's the right one.

The unit of useful agent work isn't a chat reply. It's a reviewable artifact. A chat paragraph evaporates. You nod, you scroll, and ten minutes later it's gone. A PR sits in GitHub with your name on it until you deal with it. That difference is the whole argument.

My blog bot grades its own homework before I ever see a draft. I don't let the model that wrote the post be the only judge. It sends the MDX to two cold readers from different model families, scores against a rubric, applies fixes, and loops until both say ship.

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-classifier-going-guerilla

Going Guerilla on My Inbox

What my AI inbox-screener actually does to the templated cold outreach hitting my inbox. Three real auto-sends, the footer that draws the line, and where this goes next.

zackproser.com

Just finished setting up the newsletter I've been meaning to build for three years. Get my best essays weekly, then new writing as it lands. Applied AI, dev tools, no hype. Free. zackproser.com/subscribe

Two AI agents, one phone. Claude Code handles infrastructure while Hermes runs content — fully separate, fully controllable from anywhere. Separation of concerns isn't just good software design. Turns out it's the right way to architect your agents… https://zackproser.com/blog/two-agents-one-phone

Claude Code on my Mac couldn't see what Hermes was doing on EC2. So I built a signed webhook bridge over Tailscale — SSM RunCommand triggers it, HMAC validates it, and now the two agents actually talk to each other. https://zackproser.com/blog/webhook-bridge-pattern

14,790 words. 9 posts. 47 images. 12 PRs. Two days. Full-time job. Two toddlers with zero screen time. Here's exactly how I ran two AI agents in compressed windows to make that happen. https://zackproser.com/blog/nine-posts-in-48-hours

My AI agent started giving shallow responses. So I spun up a second agent, SSH'd into the EC2 instance, found the throttled context window, fixed it, and restarted the gateway. The first agent woke up better — with no idea it happened. https://zackproser.com/blog/my-ai-agent-has-a-mechanic-agent

After weeks of frustration, I finally cracked zero-lag Steam Link streaming on Apple Vision Pro — while watching TV at the same time. Here's the exact setup that works, and every wrong turn I took getting there. https://zackproser.com/blog/play-steam-link-vision-pro-while-watching-tv

Started the day strong with an awesome workshop on Skills at @aidotengineer.bsky.social with @zackproser.bsky.social in London! If you're around the conference stop by the @workos.bsky.social booth and tell me what your favorite skill is!

9 blog posts. 14,790 words. 47 images. 12 PRs merged. All while working full-time and solo-parenting two toddlers. Zero screen time with kids. Two AI agents I talk to in 30-second bursts. One on my laptop, one on EC2. Both from my phone. zackproser.com/blog/nine-p...

I Shipped 9 Blog Posts in 48 Hours While Working Full-Time

Between a full-time job at WorkOS and solo-parenting two toddlers with zero screen time, I published 14,790 words across 9 posts in two days. 47 images, 12 PRs. The secret: chatops with two AI agents I drive in compressed windows.

zackproser.com

14 years of software engineering: RSI, neck injuries + back pain. So I stopped sitting - or even being inside. I ship to prod walking through the forest. AWS infrastructure. Real posts going live. PRs updated and merged. The phone is the new terminal. zackproser.com/blog/phones...

Phones Are the New Terminal

I gave a keynote about walking in the woods with AI. Then I stopped needing the laptop entirely. Now I direct AWS infrastructure and content agents from my phone on the trail.

zackproser.com

For me, queued messages are the biggest DevEx unlock for @cursor_ai. Now I can enter a flow state where I'm three feature requests/bug fixes ahead of Cursor. I rapidly verify functionality and put new tasks on the queue while it's working through my requests ❤️

Interest in, and adoption of, MCP is exploding. Think: one spec → any tool, any workflow, securely. Big shift for AI infra. Recap from MCPNight here: #MCP #AIInfra

Learning your stack deeply isn’t just about raw velocity—it’s about agility. When requirements shift, you can change direction gracefully, leveraging your expertise rather than scrambling for docs.

Deep expertise isn’t about being a "rockstar"—it’s about sustained rapid iteration. The tighter the feedback loop, the faster you evolve.

One thing that’s working well for Developer Education at WorkOS: pairing open-source companion repositories with detailed tutorials. It’s helping devs learn faster, solve complex problems, and build better systems. Here’s how we do it: 🧵

Hey, I’m Zack 👋 13-yr full-stack dev #buildinginpublic to help others learn. Today’s toy: an interactive token-izer—paste any text, watch an LLM slice it up (👀↓). Into no-fluff deep dives on AI/ML, RAG & #Next.js? Hit Follow and let’s geek out.

Getting deep with a single stack means fewer overlooked edge cases. You see pitfalls coming from a mile away. Production issues decrease, and deployment confidence soars.