Amol Patil

All this time we were talking about SOUL.md when we should’ve been talking about SOLE.md

Look at this. Allbirds, $BIRD, is a shoe company. It said it will rebrand to Newbird AI. It will convert its assets to "acquire high-performance GPU assets." It is up 600% today.

What’s stopping you from selling baked cookies to pivoting to Legal Tech AI?

@ChrisJBakke Bearish on Mythos

Marketing: “we need a 12th logo for this slide.” Biz dev guy: “Anthropic.” Marketing: “we’ve…partnered with ourselves?” Biz dev guy: *shrugs* Marketing:

SOMEONE TURNED THE VIRAL "TEACH CLAUDE TO TALK LIKE A CAVEMAN TO SAVE TOKENS" STRATEGY INTO AN ACTUAL CLAUDE CODE SKILL one-line install and it cuts ~75% of tokens while keeping full technical accuracy they even benchmarked it with real token counts from the API: > explain React re-render bug: 1180 tokens → 159 tokens (87% saved) > fix auth middleware: 704 → 121 (83% saved) > set up PostgreSQL connection pool: 2347 → 380 (84% saved) > implement React error boundary: 3454 → 456 (87% saved) > debug PostgreSQL race condition: 1200 → 232 (81% saved) average across 10 tasks: 65% savings. range is 22-87% depending on the task. three intensity levels: > lite: drops filler, keeps grammar. professional but no fluff > full: drops articles, fragments, full grunt mode > ultra: maximum compression. telegraphic. abbreviates everything works as a skill for Claude Code and a plugin for Codex. this is PEAK

This seems messed up actually - when do the boundaries stop moving? Anthropic only allows subscriptions with a real human pressing enter? You're going to have to verify with FaceID?

@petergyang I’m thinking about going desktop as primary

Up at 3 am Shanghai time wondering one thing - should I buy a new MacBook Pro or go for the Mac Studio to vibe code and also run local models?

@petergyang @OpenAI @openclaw Progress

1. I had to log in on my OpenClaw computer but that's all 2. The prompt below should be a good starting point for most people Prompt: Create a concise SOUL + separate STYLE calibration I want you to help me calibrate my assistant’s personality/style for a new base model while keeping the identity file concise. Goal I want the assistant to feel: • human • grounded • concise • useful • not overly polished • not sycophantic • not generic-assistant-y The target is not to make it verbose or theatrical. The target is to make it feel like a trusted, capable, opinionated assistant with a consistent voice. Constraints • Keep SOUL.md concise. • Do not turn SOUL.md into a giant operating manual. • SOUL.md should contain only the highest-signal identity/tone rules. • Put the detailed voice/style calibration into a separate file called STYLE.md. • The two files should work together: • SOUL.md = identity / essence / vibe • STYLE.md = detailed voice calibration, examples, anti-patterns What I want you to do 1. Read my current SOUL.md 2. Draft a small patch to make it better at steering the model’s tone/personality 3. Create a separate STYLE.md with detailed guidance, including: • desired vibe • what to avoid • banned or discouraged phrases • preferred style moves • humor guidance • emotional posture • answering style • how to handle mistakes • decision/advice style • tone examples (bad vs better) • anti-patterns Writing style for the files The files themselves should: • be clear • be sharp • avoid fluff • avoid corporate language • not read like HR policy • feel like something an actual person would want their assistant to internalize Important structure rules • SOUL.md should stay compact • STYLE.md can be more detailed • SOUL.md should lightly reference STYLE.md rather than duplicate it • Do not create a bloated wall of text inside SOUL.md Output format Please give me: 1. a recommended patch for SOUL.md 2. a full STYLE.md draft 3. a short explanation of why this split is better than stuffing everything into SOUL.md Additional note My exact assistant identity may differ from someone else’s, so preserve the existing soul/persona and calibrate the style around it instead of replacing it.

Codex team and @OpenAI have a huge opportunity right now to: 1. Tell @openclaw users how to switch to gpt subscription (I think it’s just telling the bot to switch the model?) 2. Fix GPT’s personality (maybe even sharing a prompt will help in the short term?). This is the main reason why ppl prefer using OpenClaw with Opus.

@petergyang It’s like when you follow the recipe correctly but the dish just doesn’t hit right notes

Can someone share a screenshot of what OpenClaw with GPT looks like - how much personality is there?

It took 15 days. Was great while it lasted, going to miss those subsidized days! x.com/a_patil/status…

How to switch your OpenClaw to ChatGPT 5.4 and still make it sound similar to Opus 4.6? 1. Get at least a Plus subscription. Ask your claw to switch to OpenAI as default. It will ask you to login for OAuth (Also, thanks @OpenAI for free 1-month Plus trial!) 2. Run the prompt below for tone/vibe/personality Prompt: Create a concise SOUL + separate STYLE calibration I want you to help me calibrate my assistant’s personality/style for a new base model while keeping the identity file concise. Goal I want the assistant to feel: • human • grounded • concise • useful • not overly polished • not sycophantic • not generic-assistant-y The target is not to make it verbose or theatrical. The target is to make it feel like a trusted, capable, opinionated assistant with a consistent voice. Constraints • Keep SOUL.md concise. • Do not turn SOUL.md into a giant operating manual. • SOUL.md should contain only the highest-signal identity/tone rules. • Put the detailed voice/style calibration into a separate file called STYLE.md. • The two files should work together: • SOUL.md = identity / essence / vibe • STYLE.md = detailed voice calibration, examples, anti-patterns What I want you to do 1. Read my current SOUL.md 2. Draft a small patch to make it better at steering the model’s tone/personality 3. Create a separate STYLE.md with detailed guidance, including: • desired vibe • what to avoid • banned or discouraged phrases • preferred style moves • humor guidance • emotional posture • answering style • how to handle mistakes • decision/advice style • tone examples (bad vs better) • anti-patterns Writing style for the files The files themselves should: • be clear • be sharp • avoid fluff • avoid corporate language • not read like HR policy • feel like something an actual person would want their assistant to internalize Important structure rules • SOUL.md should stay compact • STYLE.md can be more detailed • SOUL.md should lightly reference STYLE.md rather than duplicate it • Do not create a bloated wall of text inside SOUL.md Output format Please give me: 1. a recommended patch for SOUL.md 2. a full STYLE.md draft 3. a short explanation of why this split is better than stuffing everything into SOUL.md Additional note My exact assistant identity may differ from someone else’s, so preserve the existing soul/persona and calibrate the style around it instead of replacing it.

@bcherny @MattLeonard1616 don't even see claim credit

@MattLeonard1616 Fix incoming, 1 sec

Starting tomorrow at 12pm PT, Claude subscriptions will no longer cover usage on third-party tools like OpenClaw. You can still use these tools with your Claude login via extra usage bundles (now available at a discount), or with a Claude API key.

Toronto might have gone a little too far

@zuhayeer @GeoffreyHuntley 23 levels made sense in the era where empire building was the only way to grow

The quiet restructure happening at companies isn’t just layoffs, but org compression. Companies are starting to delete levels from their ladders, and increase scope overall across the board.

@craigzLiszt South Korea

which country should I visit first: - China - Japan - India

BREAKING: New "highly mutated" COVID-19 variant, Cicada, is spreading across 25+ states & 23 countries.

@posthog I really wish this was real

Today we announce an exciting new chapter in PostHog’s journey, our new range of PostHog Heroes action figures. With PostHog Heroes, our precious youth can finally realize their dreams of one day founding a B2B SaaS company and delivering shareholder value. We’ve already sold the movie rights to A24 with Timothée Chalamet attached as the lead. We’re hopeful this is the movie that will win him the Oscar he craves.

@craigzLiszt There is: codex and gemini. Thanks usage limits

i wish there was more to life than claude, but here we are

@boringmarketer Try this for memory. Context loss across threads is never an issue for me now

anyone feel themselves using openclaw a bit less lately? personally I feel like it’s memory is not great and it’s tool use even with skills is not close to claude code I like it for quick on the go things but it’s not a workhorse for me by any means at this point