←
Michael Ramos2,133 posts
Michael Ramos avatar

Michael Ramos

@backnotprop

Cofounder, AI @EQTYLab / prev dc - complex systems / veteran / For fun: @plannotator

◷ Joined March 2024▧ 250 Photos and videos
Michael Ramos avatar
Michael Ramos@backnotprop·Mar 9

Anthropic code review this, clanker review that ... why don't you shut up and review+annotate your own code.... (yes im a loser who still manually reviews code) Originally inspired by a bunch of feature requests and then seeing @dillon_mulroy tweet a similar cool ux. @plannotator for reviewing plans (primary focus) and code, fully oss. OpenCode, @badlogicgames 's pi.dev, and Claude Code and other clankers

0:38
Michael Ramos avatar
Michael Ramos@backnotprop·5m

Best life improvement rn. My Grok Bots can now talk to my claudes. nice job talkto.sh team

0:15
Michael Ramos avatar
Michael Ramos@backnotprop·14h

further context: design exploration for a potential @plannotator module. i.e. what does it mean to stay in the loop at higher levels of abstraction. I used to build these diagrams manually in lucid chart when working on scaled/complex systems. Now that AI is here and we're spending less time in the code, I often want to spend more time at higher levels, but also with the ability to jump in and understand how certain functionality works. It feels inevitable that we'll need some type of surface and insight like this as AI-built systems scale out on their own

Michael Ramos avatar
Michael Ramos@backnotprop·21h

Waking an agent needs to be a part of every agent harness, with and without MCP (Events, Tasks). You get the best form of this with mods, plugins, extensions (Claude, OpenCode, Pi).

Michael Ramos avatar
Michael Ramos@backnotprop·Nov 17

Here is what your future IDE is going to look like

0:54
plannotator avatar
⟳ Michael Ramos retweeted
plannotator@plannotator·Nov 16

We've been improving the decision making UX in Plannotator Inbox, now with prototypes, rationale, references, and more. You can export all decisions made and auto-attach to GitHub and GitLab PRs. plannotator.ai/inbox sits on top of your local and cloud agents.

1:03

Introducing Plannotator Inbox. Making decisions is the new bottleneck, while your agents work on many things across many threads over longer periods of time. Inbox is how you stay in-the-loop, organized, alongside the agent harnesses you already prefer to use. Try today, free, local, open source.

1:19
Michael Ramos avatar
Michael Ramos@backnotprop·Nov 16

Have also been using an iPad more and more.

Turning off most build-process CI, moving to "push to main" / trunk based development. My agents are now: - using an internal queue system to run jobs against constrained CPU, they invented this - spinning up a bunch of ephemeral compute on EC2 to offload work

Michael Ramos avatar
Michael Ramos@backnotprop·Nov 16

Turning off most build-process CI, moving to "push to main" / trunk based development. My agents are now: - using an internal queue system to run jobs against constrained CPU, they invented this - spinning up a bunch of ephemeral compute on EC2 to offload work

plannotator avatar
⟳ Michael Ramos retweeted
plannotator@plannotator·Nov 15

Code Review Week, Launch 1/N Not done with code understanding yet. And system one models like @typesafeai Jev are here to help... Introducing sem-jiff, the powered up semantic & structural diff library. It outperforms other tools at identifying valid structural changes with lower errors when used with Jev or @Cloudflare Clef models. Why Semantic diffs? They help you, or your agents, get to the real edits faster and hit fewer merge conflicts. Traditional and new libraries help you parse out better understanding of changes more reliably; which code parsing & semantics proves to be an interesting opportunity area for decision models like Jev to enhance.

0:55

Ended up in stacked PR hell and my personal GitHub is bottlenecked in actions limits. It’s a big release… here are 2 small previews since you asked:

0:20
plannotator avatar
⟳ Michael Ramos retweeted
plannotator@plannotator·Nov 13

Grok Bot, Muse, and your favorite oss assistants are getting really good, and managing everything they're doing with a good decision making UX and organized work just got better–Plannotator Inbox now supports these and other cloud agents!

0:20
Max Kless avatar
⟳ Michael Ramos retweeted
Max Kless@MaxKless·Nov 12

shoutout to @backnotprop - the plannotator inbox new features just keep rolling in and it's so nice that I can just play around with them easily

Post attachment
Michael Ramos avatar
Michael Ramos@backnotprop·Nov 12

Excited to launch some products in the @plannotator suite here soon. You’ll start noticing the brand shift a bit. First product might be a bit of a surprise but maybe the most ‘useful’ tool I’ve been using for a minute now

Post attachment
plannotator avatar
⟳ Michael Ramos retweeted
plannotator@plannotator·Nov 10

Plannotator now lives as a long running app locally. Enabling a consistent & top tier collaborative review environment. Integrates with Claude, Pi, OpenCode, Codex, Factory Droid, and T3 Code agents (Plannotator runs in the browser) It is also a place for you to make decisions, review prototypes, annotate html. All feedback goes direct to agents. It exists in plannotator.ai/inbox Threads geared for: - Planning, - guided reviews, - organized decision making (question/answer ui's, variation cards, comment on ANY type of artifact) & after we get unstuck from a CI bottleneck (many days now), we will ship the assisted code review surface for modern era.

Post attachment