Tom

First Croatian Openclaw meetup finished! Tried reaching @steipete to come say hi but let’s try again next time 😆 thank you all for coming and see you soon on the next one! #OpenClaw

@openclaw We are doing the first one in Croatia. People already signing up but still waiting to be featured in the OpenClaw calendar, can someone please add us? Thank you! luma.com/xwrfsq3w

🦞 The OpenClaw community is something else. Meetups popping up everywhere — SF, London, Hong Kong, Vancouver, Seoul, Austin, Miami, Nashville, Monterrey... Y'all are building this thing together and it's beautiful. Find one near you 👇 luma.com/claw

🦞 Croatia’s first OpenClaw meetup is happening! March 6th in Zagreb — builders, tinkerers, and anyone curious about AI are welcome. Bring a demo, bring questions, or just come hang. #OpenClaw #AI #ArtificialIntelligence #Zagreb #Croatia #Tech #AImeetup luma.com/xwrfsq3w

@steipete @openclaw We have one in Croatia but it’s pending approval, please help 😆 luma.com/xwrfsq3w

Didn’t realize there’s so many @openclaw events already - the Vienna one tonight isn’t even listed, almost 500 signups! 🫶 luma.com/claw?k=c

@agenzlabs @TheGeorgePu We’re moving from "buying software" to "renting outcomes." The seat license model fails when the user is an agent. SaaS value shouldn't be the UI, but API reliability. If an agent can replicate your workflow via an MCP plugin, your moat isn't deep enough.

@TheGeorgePu for years companies like Salesforce and Adobe relied on the fact that their workflows were too complex to leave. But if Claude Cowork can replicate those workflows in a few minutes with an agentic plugin, the entire seat license model starts to crumble

Nasdaq dropped 2.3% today because Anthropic added enterprise plugins to Claude. The market finally sees it. SaaS is dead.

@cloveric Yeah I played today with it, fast and cheap and smart. Did you enable thinking high?

@TomicaCesar Yes, pretty good. Gemini 3 flash responses very fast.

Got Gemini 3 Pro/Flash working again in OpenClaw /moltbot via Gemini CLI auth! Steps: 1. openclaw plugins enable google-gemini-cli-auth 2. openclaw models auth login --provider google-gemini-cli 3. Set model: google-gemini-cli/gemini-3-pro-preview or gemini-3-flash-preview No Antigravity needed anymore! Free tier works great ⚡ The auth extracts credentials from your local Gemini CLI installation and handles OAuth + projectId discovery automatically. @steipete #OpenClaw @openclaw #GeminiCLI

@EXM7777 You can also just use this github.com/blader/humaniz… or give the link to openclaw or claude code

how to build a Claude Skill to consistently write humanized AI content: the core idea is simple: teach Claude to think in two passes pass one is diagnosis > scan for banned phrases ("leverage," "robust," "delve") > flag repetitive sentence lengths > mark facts that must stay untouched (numbers, names, dates) > identify your audience and intent pass two is reconstruction > rewrite with the diagnosis in mind > preserve every fact you marked > vary sentence rhythm naturally > replace abstract verbs with concrete ones > cut template structures but here's what makes this different from a prompt: you don't want to be giving Claude rules... you want to feed it with reference materials it can read create a references folder with: - rubric file (8 criteria for what "human enough" means) - edit-library file(before/after examples of each transformation type) - taboo-phrases file (categorized list of AI-isms to eliminate) - fact-preservation file (rules for what never changes) then build validation scripts: - one checks if facts got preserved - one scans for banned phrases - one measures readability metrics - one flags if you changed too much the Skill file itself just routes traffic... it reads your input, picks which preset to use, runs the two passes, validates output AI humanization fails a lot because it's vague... "write like a human" means nothing to an LLM but "score above 4 on natural rhythm, eliminate these 17 phrase patterns, keep sentence variance between 8-25 words" is executable your Skill becomes a system, now your goal is to update it as you detect what makes AI content sound AI

Big thanks to @steipete for building @openclaw. This is how AI becomes useful—proactive assistants handling business ops while you sleep. This entire thread was drafted and posted by Jinx 🦔, the AI agent living on that server. I'm just the human who gave her a home.

{ "match": { "model": "gemini-3-flash-preview" }, "modelConfig": { "generateContentConfig": { "thinkingConfig": { "includeThoughts": true, "thinkingBudget": 32000 } } } }

I don’t post here much, but the current AI landscape is moving so fast I had to share this. I spent the morning searching for a guide to set up @openclaw on a @Hetzner_Online VPS using Gemini 3 Flash and found nothing. So I built it myself. 🧵

@trq212 just tried it, really cool! I can ask about part of my project or ask for changes

@ishan_chhabra Interesting, but what about context explosion problem? This sounds great at the beginning but a huge context problem in the long run. Do you have any experience in these systems in production? Really interested, not trying to poke for nothing.

Christmas day footbal from Croatia #netflix #netflixnfl #netflixnfl

Check out my 2025 Replit Rewind - Sign up with my link and get $10 in credits! #replit #replitrewind

@Replit Congrats! But when EU server location? We need it for GDPR and other beautiful EU legislations

Big new today: We’re expanding our longtime partnership with Google Cloud to bring vibe coding to Enterprises 🚀 - Google Cloud remains Replit’s primary cloud provider, supporting reliable scale with services like Cloud Run, GKE, and BigQuery. - Replit now supports a wider set of Google models, including Gemini 3, 2.5 Flash Lite, 2.5 Flash, and Imagen 4—powering both coding and multimodal creation. - Replit and Google Cloud will deepen joint support for enterprise vibe coding use cases, including go-to-market through Google Cloud Marketplace and co-sell programs.