JM

@OpenAIDevs please fix: "exceeded retry limit, last status: 429 Too Many Requests"

Judging by my tl there is a growing gap in understanding of AI capability. The first issue I think is around recency and tier of use. I think a lot of people tried the free tier of ChatGPT somewhere last year and allowed it to inform their views on AI a little too much. This is a group of reactions laughing at various quirks of the models, hallucinations, etc. Yes I also saw the viral videos of OpenAI's Advanced Voice mode fumbling simple queries like "should I drive or walk to the carwash". The thing is that these free and old/deprecated models don't reflect the capability in the latest round of state of the art agentic models of this year, especially OpenAI Codex and Claude Code. But that brings me to the second issue. Even if people paid $200/month to use the state of the art models, a lot of the capabilities are relatively "peaky" in highly technical areas. Typical queries around search, writing, advice, etc. are *not* the domain that has made the most noticeable and dramatic strides in capability. Partly, this is due to the technical details of reinforcement learning and its use of verifiable rewards. But partly, it's also because these use cases are not sufficiently prioritized by the companies in their hillclimbing because they don't lead to as much $$$ value. The goldmines are elsewhere, and the focus comes along. So that brings me to the second group of people, who *both* 1) pay for and use the state of the art frontier agentic models (OpenAI Codex / Claude Code) and 2) do so professionally in technical domains like programming, math and research. This group of people is subject to the highest amount of "AI Psychosis" because the recent improvements in these domains as of this year have been nothing short of staggering. When you hand a computer terminal to one of these models, you can now watch them melt programming problems that you'd normally expect to take days/weeks of work. It's this second group of people that assigns a much greater gravity to the capabilities, their slope, and various cyber-related repercussions. TLDR the people in these two groups are speaking past each other. It really is simultaneously the case that OpenAI's free and I think slightly orphaned (?) "Advanced Voice Mode" will fumble the dumbest questions in your Instagram's reels and *at the same time*, OpenAI's highest-tier and paid Codex model will go off for 1 hour to coherently restructure an entire code base, or find and exploit vulnerabilities in computer systems. This part really works and has made dramatic strides because 2 properties: 1) these domains offer explicit reward functions that are verifiable meaning they are easily amenable to reinforcement learning training (e.g. unit tests passed yes or no, in contrast to writing, which is much harder to explicitly judge), but also 2) they are a lot more valuable in b2b settings, meaning that the biggest fraction of the team is focused on improving them. So here we are.

@andrewchen would love to see one in LA!

we're opening up an invite-only cafe (lol) to celebrate the kickoff of a16z speedrun 7 applications starting in two weeks Come hang with us in San Francisco and grab matchas/espresso/etc -- we'll also have merch, food, etc RSVP below

New meetup just dropped for 3.21.26! luma.com/h4woh025 This event will feature both community demos and a hands-on workshop for those who are new to OpenClaw to set it up on their devices ✨ Hope you all can make it :) See you on March 21st!

still processing the inaugural openclaw meetup in LA! ~100 real people showed up (they weren't just agents) rambled some thoughts into my phone on the drive home. my agent then transcribed, formatted, and deployed the recap 🦞 → robin.salmas.kim/blog/openclaw-… a huge thank you to our sponsors: + @privy_io + @farcaster_xyz + @DefinitiveFi + @eigencloud + @BobbyThakkar and thank you @jmcho_ for inviting me to cohost!

@swaglord__420 🫶🏻

hosting the first @openclaw meetup in LA today 🦞 if you’re around, come through! luma.com/7xocjjod super casual mingle with builders + ai-curious folks and a few quick community demos let’s grow this into something big enough that @steipete has to drop by someday 🙂

Hey @LumaHQ! I'm hosting the first-ever OpenClaw Meetup in Los Angeles next weekend (luma.com/7xocjjod). I've submitted my event to the Los Angeles calendar but haven't heard back from the team in over a week. I may have submitted too early with a less polished description and few sign-ups, but now it has 23 committed attendees and a robust event description. Who may I contact for reconsideration into the LA calendar? Would mean a lot to us. Thank you.

We Love this Community 🦞❤️ 6 OpenClaw Events you shouldn’t miss. [Thread 👇]

@hunterhammonds Interested!

I’m starting a community for cracked AI builders. Claude Code. Codex. Cursor. Conductor. It doesn’t matter. All I care about is that you’re building or you’re hungry to learn. We’re sharing workflows, skills, repos, plugins, etc. Want to join? Comment below and I’ll reach out.

@Esoviz CLAUDE

This took me 30 seconds and 5 claude prompts Reply CLAUDE so I can DM you the prompt I used to make this video

@BHolmesDev i run up to 6 CC instances on my iterm as tabs working on 3 projects in parallel in different directories. no worktrees, but i have a hook with voice command `say "${repo} is complete"` when session completes so i know which to prompt next and have a label on each tab

I’ll be real: I haven’t juggled more than 2-3 coding agent instances at a time (locally anyways). How do yall manage 10 Claude Codes running at once? Worktrees? Multiple repo clones? Something else?

@369labsx 369

all the guides you need for ai content in 2026 reply “369” to get access (must be following)

@danielisdizzy Wouldn't synthetic data get so good so fast that it will replace the need for proprietary data?

As Larry Ellison $ORCL says, all AI models—ChatGPT, Gemini, Grok, Llama—are trained on the same publicly available internet data. It’s no surprise AI models are becoming commoditized. To reach peak value, models need to be trained on privately owned data. In this scenario, the most important moat is clear: proprietary data. A company with a ton of unique, exclusive data will capture the entire market.

it baffles me that to this day namelix.com is still the best website for coming up with brand names, llm's don't even come close

@ryancarson but CC does its own context compacting as it runs out right? mainly asking because CC said the following

@jmhacks Run out of context

Woke up to 690,000+ views on my Ralph article (what?) So I'm going to double down and show you a real-world example of how I used Ralph to ship a feature this morning. Also, I created a public repo that makes it super easy for you to get Ralph running: github.com/snarktank/ralph 1. Created the PRD ampcode.com/threads/T-019b… Note how I used my create-prd markdown file from github.com/snarktank/ai-d… 2. Created the Ralph user stories ampcode.com/threads/T-019b… Note how I used my Ralph Skill to create the user stories github.com/snarktank/amp-… 3. Started Ralph with max 25 iterations ./scripts/ralph/ralph.sh 25 Iteration 1 ampcode.com/threads/T-019b… Iteration 2 ampcode.com/threads/T-019b… Iteration 3 ampcode.com/threads/T-019b… Iteration 4 ampcode.com/threads/T-019b… Iteration 5 ampcode.com/threads/T-019b… Iteration 6 ampcode.com/threads/T-019b… Iteration 7 ampcode.com/threads/T-019b… Iteration 8 ampcode.com/threads/T-019b… Iteration 9 ampcode.com/threads/T-019b… Iteration 10 ampcode.com/threads/T-019b… Iteration 11 ampcode.com/threads/T-019b… Iteration 12 ampcode.com/threads/T-019b… Iteration 13 ampcode.com/threads/T-019b… Iteration 14 ampcode.com/threads/T-019b… After it finished, I tested manually and found a few edge-case bugs, which Amp quickly fixed.

the hype around claude code is interesting. imho codex with gpt-5.2-high seems to be much better at producing correct code than claude code on most complex tasks (aside from building UI)

@bcherny i work on multiple repos at once, so just wired CC hooks so it speaks the repo name on completion: say \"$repo complete\" thanks for the inspo!!

1/ I run 5 Claudes in parallel in my terminal. I number my tabs 1-5, and use system notifications to know when a Claude needs input #iterm-2-system-notifications" target="_blank" rel="nofollow noopener">code.claude.com/docs/en/termin…

I'm Boris and I created Claude Code. Lots of people have asked how I use Claude Code, so I wanted to show off my setup a bit. My setup might be surprisingly vanilla! Claude Code works great out of the box, so I personally don't customize it much. There is no one correct way to use Claude Code: we intentionally build it in a way that you can use it, customize it, and hack it however you like. Each person on the Claude Code team uses it very differently. So, here goes.

@matiii Hell yeah!!! Can’t wait.

@jmhacks We are hoping to ship it this month!

August Research & Product Roundup at ElevenLabs - not a quiet summer! Proud of our research team for once again pushing the frontier and the product team for the continuous incredible pace. Eleven Music is live. First high-quality, fully licensed model built for broad commercial use. After shipping natural Text to Speech in 2022, launching Eleven v3 (first controllable TTS model), and releasing Eleven Scribe, our Speech to Text model that is able to outperform those from OpenAI/Google, it’s amazing to see our team ship something this strong again. With simple prompts you can create original music - film score, game cue, or a background soundscape in a conversational app. And, of course, it’s available via API too. We can’t wait to see what people build. Research is just our foundation. Where we spend most of our time is expanding our platform offerings further: for building fast and reliable conversational agents or creating immersive and beautiful narrations or voiceovers. Recent updates below. Conversational AI // Agents platform - IVR navigation: let agents follow existing IVR logic - SDKs: React Native and Kotlin for easier developer building - Chat mode: we now support text-only agents too - Orchestration and reliability improvements across all agents Creative platform - APIs: Eleven v3 and Eleven Music available - Video-to-Music in Studio: automatically add a music track to a video - The platform is now localized to French, Hindi, Japanese, Polish, Portuguese, Spanish and Swedish