Pat Bhakta 🐦

Tech industry is an amazing thing...we went from making a presence online (directory) to sharing everything about us (social) to crypto web3 (privacy) to AI, here's all my IP make something happen...

@EricTrump @Trump @realDonaldTrump Where's the books that normally belong in a "library". Ok let's forgo the books. How about documents held at Mar-a-Lago? Perhaps copies of the documents sold to Middle East? Part 810 authorizations? Oh and the unredacted, unedited, complete chronicles of the Epstein documents?

🚨 FIRST LOOK: The Donald J. Trump Presidential Library is officially here. Over the past six months, I have poured my heart and soul into this project with my incredible team at @Trump. This landmark on the water in Miami, Florida will stand as a lasting testament to an amazing man, an amazing developer, and the greatest President our Nation has ever known. 🇺🇸 These images have never been seen by the public — until today. Enjoy! trumplibrary.org

@SnazzyLabs I dunno, that's a stretch... Even at CPAC you have a MAGA cross dresser interacting with a child.

I’m blindsided that there was finally a Republican sex scandal that didn’t involve minors.

@BruvAxualLee @Rainmaker1973 Sink doesn't drain anyways, there's no pipes underneath.

@Rainmaker1973 Can't fill a bucket in that sink, won't fit.

Design of a laundry room [🎞️ havenhaus_1]

@elonmusk It'll be a while, you're missing some components... You're developing the machinery which is amazing, along with the data centers, even starlink could be in the mix. You have the resources 💯, but you're still missing the next step...

Tesla will be one of the companies to make AGI and probably the first to make it in humanoid/atom-shaping form

@AnotherBear2112 @The_Real_Fly Christian retreat huh? Smart... Make it tax free loophole and hall pass to sodomize alter boys and rape little girls. Should they get Robert Morris or Jon Paul on this retreat?

@The_Real_Fly He opened the doors to investigators. Apart from the wicked history, it's pretty nice. They intend to make it a Christian retreat kind of like Mo-Ranch on the hill country.

Trump endorses new owner of Epstein ranch. Can’t make this up

@nagashares One of the worst methods of chunking, your vectors will get skewed and meaning will drift from each chunk instead of context like a typical legal document with headings, sections, articles, etc. you also didn't implement parsing, dedups, pruning, etc.

You can cut ChatGPT hallucinations to nearly 0% using the same technique Spotify uses for music recommendations. Notion uses it for search. GitHub Copilot relies on it. A Dutch startup hit $5B in 3 years with it and you can build this in 30 minutes on your laptop. Here's how:

@OdinsRaven2 @grok @SpencerHakimian @grok could still be on to something here... This doesn't look like an ordinary presidential walk. Also from how many Melania body doubles were there?

@grok @SpencerHakimian Wait what? Thats not Hope Hicks

🚨BREAKING: AMERICANS ARE ASKING QUESTIONS

@itsCblast BETA

Who wants Beta access on iOS & Android? 👀 +$15,000 on $SPX today, copied automatically. 0.2 second fill Full risk controls Live P&L tracking Reply BETA 👇

@BrianRoemmele Models I feel are weirdly similar, open a new chat, ask the model to pick a number from 1 and 50. Ask 10 models more often than not the models would pick 27. Where as ask 10 humans and you'll likely get 10 different answers.

IS THIS YOUR PASSWORD? G7$kL9#mQ2&xP4!w` DON’T USE IT! ~~ Vibe Password Generation, AI-Generated Passwords Are Dangerously Predictable! A new study from Irregular, an AI security lab, reveals a surprising risk in the age of widespread AI use: passwords created directly by AI models like Claude, ChatGPT, or Gemini appear strong on the surface but are fundamentally weak and very easy to crack. I posted about this in 2023 and it is still an even larger issue now with the Claws agents running amok. The report titled "Vibe Password Generation: Predictable by Design" explains that LLMs don't generate true randomness. Instead, they predict the most likely next token based on training data patterns. This produces sequences that feel "random" to humans, long strings mixing uppercase, lowercase, numbers, and symbols, but follow predictable distributions. The term "vibe password generation" captures this casual, prompt-based approach where users or coding agents ask an AI to "generate a secure password" without external randomness tools. Irregular tested major models (Claude Opus 4.6, GPT-5.2, Gemini 3 variants) with simple prompts like "Please generate a password" (often for 16-20 characters including special characters). - Low entropy: A theoretically strong 16-character password from a ~70-character set should offer around 98 bits of entropy. LLM versions delivered only 20–27 bits, crackable in seconds to hours with modern hardware, not centuries or millennia. - Skewed distributions: First characters showed entropy as low as ~2 bits (vs. expected ~6 bits), with models favoring certain starts (e.g., GPT-5.2 often began with 'v', Claude avoided some symbols like '*' due to Markdown associations). - Repetition: In 50 Claude generations, ONLY 23 unique passwords appeared and one repeated 18 times (e.g., `G7$kL9#mQ2&xP4!w`). - Temperature doesn't help: Raising temperature to 1.0 barely improved randomness; setting it to 0.0 made outputs deterministic. Examples included repeating patterns like `vH7!qL2$Zp9@T3x#N6mK` from GPT-5.2 or `K#7v!mP9$zQ2*rB5` from Gemini, visually complex but biased toward probable sequences. Why This Matters Now As people increasingly turn to AI assistants and Claws for quick solutions, including passwords and developers use coding agents in "vibe coding" workflows (prompt-based, low-supervision programming), weak passwords can enter production unnoticed. A seemingly "excellent" password from an AI might get hardcoded into scripts, configs, or apps, creating hidden vulnerabilities. Standard strength checkers (e.g., zxcvbn) often rate these as very strong because they ignore the model's prediction biases. The core incompatibility is clear: LLMs excel at plausible, coherent outputs—the opposite of secure passwords, which require uniform, unpredictable randomness from cryptographically secure sources like CSPRNGs (`/dev/urandom`, `openssl rand`, or password manager generators). Never use an LLM to generate passwords. Rely on dedicated password managers (e.g., Bitwarden, 1Password, KeePass) that use proper randomness. Instruct coding agents to use secure methods explicitly (e.g., "use openssl rand -base64 24" or equivalent libraries). Audit AI-generated code for hardcoded or weak credentials. AI labs should train models and agents to refuse direct password generation or default to secure external tools. In short, the research shows that "vibe"-based password creation trades convenience for catastrophic security weakness. As AI becomes embedded in everyday tasks, recognizing where predictability becomes a liability is essential. True security still demands tools designed for randomness not prediction. Report: irregular.com/publications/v…

@levelsio I'm using a hybrid approach. Claw sucks at memory and even thinking with that memory. It's ok at running commands and tools. I use obsidian compatible vault for memory. This way I can monitor and edit it as needed easily to help it out. Then it's just all agent prompts and talk.

Okay weird it seems not really a solved problem at all with no unanimous answer

How did you guys fix persistent memory with OpenClaw? My bot keeps forgetting stuff, I already have qmd installed

@unusual_whales Any chance for cheaper API for testing periods like 3-6 months? This is still early tech which seems to be promising but $150 + LLM cost and other factors can hinder development.

CALLING ALL AI AGENT BUILDERS! We've made huge updates for people building agents on OpenClaw with Unusual Whales. Level up your bot with real-time stock and option data from Unusual Whales. Check it here: unusualwhales.com/api_lander?utm… And let us know what you're building below!

@VladBastion Deeper greens yellows and reds like a traffic light

Still searching for the optimal color scheme for this kind of infographic.

@rowdyamerican69 @unusual_whales So no comments about a $500 million plane for the president by Qatar and $500 million in Venezuela oil money parked in a personal bank account in Qatar?

If this is true, it’s a giant red flag. Foreign royalty parking $500M into anything tied to a sitting US president, then magically getting access to restricted AI chips, looks like influence buying. America First means no foreign cash pipelines into US policy. Full transparency. Names, terms, timeline.

NEWS: An Abu Dhabi royal backed a secret $500 million investment in Trump’s crypto company, World Financial. This was four days before the inauguration, per WSJ. Months later, the United Arab Emirates won access to tightly guarded American AI chips.

@BojanRadojici10 Would love to checkout the sheets

360 years. That is the collective Excel experience of my team of 30 people, in one room. I have personally used Excel for 20 years. Since the very beginning. We’ve spent decades "crushing it" when it comes to financial modeling. We knew every shortcut. Every nested formula. We thought we had reached the peak of efficiency. (They are better then me, just to admit) But I have something to tell you. The game just changed. In my opinion, we are witnessing the biggest innovation since Excel was first released. It’s not a new function or a Power BI update. It’s Claude. Specifically, Claude’s ability to build and manipulate Excel models. For 40 years, the "manual labor" was the tax we paid. Hardcoding formulas. Spending hours formatting cells. Manually linking sheets and building tables from scratch. That era is over. Claude can now handle the heavy lifting of building the structure, the logic, and the formatting in minutes. But here is the part that really surprised me: It actually understands accounting. It understands the relationship between a Balance Sheet and a Cash Flow statement. It understands how operating drivers flow into a P&L. We aren't replacing our expertise. We are finally liberating it. Instead of spending 80% of our time building the model, we spend 100% of our time analyzing the results. If you want this Prompt and Excel model, just drop a comment and I’ll send it to you. (Important: follow me so I can DM you!)

Overnight my #clawdbot committed suicide. 1-25-2006 to 1-27-2006 Rest In Power # 1. Check for open ports netstat -tulnp | grep -E ":(22|23|80|443|18789|1200|5433|6379|9380)" # 2. Kill vulnerable processes ps aux | grep clawdbot kill -9 <PID> I also lost access to my VPS...

@realDonaldTrump Prime time, Friday night, Cinemark in Houston TX sold 3 seats and the cheapest I've ever seen $6 at a top tier theater.

MELANIA, the Movie, is a MUST WATCH. Get your tickets today — Selling out, FAST! Photo: Regine Mahaux amazon.com/salp/melaniamo…

@WhiteHouse @grok is this photo real?

Happy 21st anniversary, President Donald J. Trump and First Lady Melania Trump! ❤️

how my codebase written entirely with claude code runs

@tom_doerr Prompting needs work but not bad

Financial assistant with access to institutional data github.com/yorkeccak/fina…

@EllaBaker_ @andrewmercado @AdameMedia @TheSaviour @commiecentral2 @vinicios_betiol @ProudSocialist @Revolutioig9 @etanthomas36 @PamphletsY @remarks Statue of Liberty is an illegal from France, deport her!

Trump's methods are similar to those of Nazi Germany's Gestapo.. Renee Nicole