jpreagan
Re:How I Learned to Stop Worrying and Love OpenClaw
on Friday February 13, 2026 @02:26AM (#46999912)
Great question. I touched on this in the article: "our assistants will deal with one another, taking action on our behalf."
The honest answer is nobody knows exactly what that looks like at scale. Markets with millions of automated agents optimizing simultaneously could behave very differently. That's pretty wild to think about.
But I think the more immediate reality is much more mundane. Right now these assistants are reading your email and reminding you about dentist appointments. We're a long way from autonomous agents negotiating on your behalf at scale. The infrastructure isn't there, the trust isn't there, and the coordination protocols between agents don't really exist yet. When we do get there, it probably looks less like a free-for-all and more like new norms and rate limits emerging, the same way APIs and marketplaces already throttle automated access. That's my guess.
The interesting question to me is who builds those protocols and whether they're open or proprietary.
Re:How I Learned to Stop Worrying and Love OpenClaw
on Friday February 13, 2026 @01:20AM (#46999522)
I built a personal AI assistant on a Mac mini with OpenClaw. Two-machine architecture, zero public exposure via Tailscale, read-only message bridge, and dedicated Apple ID separation. Here's how I set it up and why I think personal assistants are the most exciting thing happening in AI right now.
Show HN: llmnop – Rust CLI for benchmarking LLM endpoints
https://github.com/jpreagan/llmnop
I built this because I needed to benchmark LLM inference endpoints and the existing tools required Python environments. I wanted a single binary I could grab quickly on any server.I've also become interested in performance metrics like time to first token, inter-token latency, throughput, and wanted a tool focused on just that.llmnop is written in Rust and was initially modeled after LLMPerf,...
Re:Llmnop, a tiny Rust rewrite of LLMPerf
on Monday July 14, 2025 @05:20PM (#44565478)
I've been using LLMPerf for a while to evaluate the performance of our inference servers (vLLM, SGLang, etc.).
It works great, but I was running into memory constraints while testing large number of concurrent users on some servers, and didn't always find the specific Python version requirements convenient.
So, I rewrote the benchmark aspect of this tool in Rust to get an easy single-line install.
I hope its useful to others as well, and would love to hear feedback if you have any suggestions for improvement.