←
Rich Steinmetz1,200 posts
Rich Steinmetz avatar

Rich Steinmetz

@RichStoneIO

Weeklyish AI Engineering nugget: looplabs.cc/newsletter 🧪 🚂 Organizing softwaremakers.ai & Rails.Builders 🤖 AI navigation at @ClickFunnels since 2023

◷ Joined October 2012▧ 147 Photos and videos
Rich Steinmetz avatar
Rich Steinmetz@RichStoneIO·Nov 8

Coding by hand today for the first time in 2026 on a Sunday morning. It's surprisingly bizarre and fun. Feel free to code review. 🤣

Post attachment
Igor Aleksandrov avatar
⟳ Rich Steinmetz retweeted
Igor Aleksandrov@igor_alexandrov·Nov 6

Tabletist just passed 50 stars on GitHub. Thank you to everyone who tried it and sent feedback. Next goal: Homebrew's official casks. They take an app at 225 stars when its author submits it, so if Tabletist is useful to you, a star helps. github.com/igor-alexandrov/t…

Rich Steinmetz avatar
Rich Steinmetz@RichStoneIO·Nov 4

Ooops, I think I automated myself out of an amazing 3.5-year ride at ClickFunnels. 😅 Today was my last day and, as with all endings, my first question is: what's next? Which AI-centered team needs someone who enjoys product work, community work, wears different hats (sometimes quite literally), can still read PRs and assess them with a sharp product focus, but lets AI write (and, frankly, often rewrite) the code? If that sounds like your team, or you know one, I’d love an intro :)

Post attachment
Jason Sich avatar
⟳ Rich Steinmetz retweeted
Jason Sich@jasich·Nov 1

Last week I launched an AI support agent for Overskill named Riley. 6 months ago “good” would have been the equivalent of level 1 support. Riley has the title VP of Customer Support. “She” will be the entire department.

This is one of the better litmus tests for whether you’re pushing your use of AI enough or not. Up until maybe six months ago probably the best you could do with AI was pull off something you already would produce, but just doing it a little faster. But now we’re entering a point where the output of working with AI can and should dramatically exceed what you would been able to do before, in a fraction of the time. The product prototype should not only be fully functional and interactive, but it should probably have meaningfully better features than it would have had before. The engineering architecture should be able to deliver performance gains that you wouldn’t have been able to go after otherwise. The research for a project should consume every single fact on the subject across every dimension of analysis possible. The business process you implement with AI should operate entirely differently and learn from all its past failures in a way no human could. We’re unquestionably entering a period where AI will be smarter than us on most axes. Perhaps not on every dimension that matters (intelligence is still very jagged), but at least on enough dimensions that when we combine it with what we’re good at, you get qualitatively different outcomes. So the simple test is: if you could have done what you’re doing before AI, you’re likely not pushing it enough.

Rich Steinmetz avatar
Rich Steinmetz@RichStoneIO·Oct 24

Why video QA is better than screenshot-based or text-report QA. True story, the bots are gonna lie to you. It's gotten much better over the months and years, but it's still true.

Post attachment
Rich Steinmetz avatar
Rich Steinmetz@RichStoneIO·Oct 23

If you are one of those engineers who are craving back the art of engineering now that you don't write code anymore (and don't care about it for certain types of products), owning your agent through Pi could restore the spirits. Pi Durable is the next evolution of this: It's separate from the Pi harness right now, but gives your agents a persistence layer to be run (and killed) anywhere anytime. youtube.com/watch?v=DNDYYHsy…

Jason Sich avatar
⟳ Rich Steinmetz retweeted
Jason Sich@jasich·Oct 22

I have over 100 agent sessions running locally. Today I gave them all an MCP to reach me over Telegram. Feels like publishing my phone number on the internet for anyone to call.

Rich Steinmetz avatar
Rich Steinmetz@RichStoneIO·Oct 22

If you are tired of your brain and you would like to fry it one last time, start a few parallel /grill-me sessions. Thank me later.

Post attachment
Rich Steinmetz avatar
Rich Steinmetz@RichStoneIO·Oct 21

I've been paying OpenAI $100-$200 for Codex over the past few months and seem to have accumulated ~$2,100 in credits, which is cool, plus all the resets I'm getting during the month, like 5-6 at least. So, not complaining, but I noticed that tokens started melting seemingly faster yesterday. Time to pull some agent analytics and see hard proof of what's going on.

Post attachment
Rich Steinmetz avatar
Rich Steinmetz@RichStoneIO·Oct 19

I have a bunch of commands that start as a Raycast keyboard shortcut to automatically paste a text snippet I use frequently. If I need to type the same thing 2-3 times, it becomes a Raycast snippet. If I keep using that snippet frequently, it might become a skill or a memory. That way, I don't spam my skills too much and still give enough room for automation before graduating an idea to something the bots will be able to use autonomously or on demand later.