Plan mode is soooo back

best use case for jev is a2a

Jev by @typesafeai turns your  Mac into an AI agent... what this means for You ? - Jev, a new model from a ChatGPT co-inventor, can operate computer tools in real time. - It is 200x faster for Computer use and 400x cheaper in addition to the optimizations. - Yappy from @MitosisLabs is an AI that lives on your mac and do several things in parallel. Filling out Job applications, doing your Taxes and other documents, filling out forms, and so on... - With Jev, all these things happen at the blink of your eye! Below is a video demonstrating this IRL. Here are the benchmarks (Yappy + Jev V/S our competitor): - End-to-end time: 1m 54s vs 13m 14s - LLM Calls: 22 vs 104 - Input tokens consumed: 0.39M vs 11.24M - Estimated cost of the run: ~$0.24 vs ~$4.24 Link to download in comments.

i don't want AI in your app i want your app in my AI

openclaw really died because the founder had no aura

if your startup is not agent ready you are ngmi

My friend is the highest paid developer in Berlin He just got promoted to the CTO position His new salary? A jaw-dropping €51,000 a year We went to the bar to celebrate his promotion “So how much is it after taxes?” I asked him “€23,000 a year” he said proudly Suddenly, the American tourist sitting next to us obnoxiously shouted “Sorry, but are you guys serious? This is what a cleaning lady makes in the US, not an engineer salary lol” My friend and I look at each other in disbelief First of all, €51,000 is an extremely high salary for a developer in Germany Secondly, unlike Americans, he will never be fired, so he can earn €23,000 for the rest of his life But you can't expect a yankee to understand all that No wonder America is collapsing if they keep paying their engineers unsustainable wages

the deepseek harness is the best harness

Someone put the before and after @thsottiaux in the same room. This is better than my reply

This is the agentic room. The Agent Native Builders Hackathon runs Wednesday 26 and Thursday 27 August at @Cloudflare's San Francisco office. Two days, and one question underneath all of it: can an agent that nobody briefed arrive at your product cold, work out what it does, use it, and pay for it. Most software answers no today. It was built for a person who already knows where the buttons are. The builders arriving tomorrow are not starting something new for a weekend. They are bringing things they already ship and opening them to agents, which is harder. With $30K+ pool of builder credits and prizes! There are only 50 seats. Organised by @immersivecommon , @RuntypeLabs , Nate Stewart Nathan Booker and Cotal.ai Sven Jonscher David Farah. Co-hosted by Cloudflare, thank you for the space! @nebiusai Sujee Maniyam, co-host and sponsor. Special thanks to our sponsors @tenki Colin Lowenberg Guzman Pintos, @tavilyai Greta Ernst, @AIsaOneHQ ., @hud_evals Axel Cassou, @MitosisLabs Prakshal Jain Alex Morris, Whiterabbit and Hacker bob. With community partners @agentcommunity_ Balázs Némethi, Learning Layer Labs Devinder Sodhi, @ClawCampus Steven Echtman Soji lee, @HireyAI Walter Wu Cheryl Yang and @nerdconf_ar Alex Marin. Judged on Thursday 27 August by Michalis Vasileiadis, Axel Cassou, Barton Rhodes, Fabio Roma and Vidya Bodepudi. There are 10 spots reserved for wildcard entries! Apply! lnkd.in/gBreBNag hashtag#AgentNatives hashtag#AIAgents hashtag#AgenticAI hashtag#Hackathon hashtag#SanFrancisco

Hacker Bob is a judge and sponsoring the Agent Native Builders Hackathon. Security testing for web apps, smart contracts and source code. 17 CVEs across 9 projects came out of this repo. @memehalis Aug 26-27, Cloudflare SF. luma.com/agentnativebuilders… #AgentNatives #AIAgents

We are organizing the Agent Natives Builders Hackathon for people already building products and startups that are already shipping. When: August 26 and 27 Where: @Cloudflare's San Francisco office Fifty seats only. Over the course of the year, our organizers and mentors have built top ranked agent-ready platforms and shipped foundational agent-native tech. This is a rare chance to build alongside them. Bring a product and leave with an agent surface that works: discovery, authentication, tooling, payments, usability and accessibility. We have two tracks. - External. What your customers' agents can do with you. Your product, reachable by someone else's agent. - Internal. What your own team's agents can do for you. The work behind the product, run by agents you trust. This hackathon will test your product across the full stack of agent-first tech infrastructure. Solo builders are welcome. You will work with our partners' forward deployed agents, hosted on Immersive Commons agent-native infrastructure. RSVP: luma.com/agentnativebuilders… Hosted by @Cloudflare SF Organized by @RuntypeLabs , Cotal.ai and @immersivecommon Sponsors and partners: @tenki, @nebiusai, @tavilyai, @MitosisLabs and Hacker Bob Community partners: @agentcommunity_, Learning Layer Labs, @ClawCampus , @dablclub, Web A and @HireyAI

has anyone tried glm 5.3?

today I installed windows on two machines. first thing I downloaded was the chatgpt app to do every other configuration through it. chatgpt is the new super app

I love the DeepSeek harness's vision

Exposing the SECRET of Wispr Flow 🔎😱 It reads what's on your screen (Bank account pages, Private Chats) and sends it to servers that aren't even theirs. On by default. Wispr Flow is the most popular dictation app on the market: 4,000+ businesses, millions of users, and $260M raised last month. You don't hand $260M to a voice-to-text app you could build in a weekend. So I got curious and reverse-engineered their flagship product to find out what's really going on. 1. Every time you dictate, Wispr reads the text around your cursor: your open DMs, your bank balance, the doc behind the window, plus your app and the exact URL you're on. And... Sends it to their cloud for inference. Have your bank account open, a private DM, or a confidential doc open behind the cursor? That context goes too. This is ON by default. It's controlled by a flag called useAxContext `(default: true)`. not by any privacy setting you can see. And full-screen capture? It's already wired into the shipped app `(shouldOCRScreenCapture, desktopCapturer.getSources({types:["screen"]})` grabs your entire display, not just the active window). It's one toggle away, sitting in the code, ready to go. 2. None of this runs on your computer. It runs on servers that aren't even Wispr's. Open Activity Monitor while you dictate and watch the connections light up. Your audio, plus everything above, gets sent to Baseten, a completely separate third-party company, for processing, then to Wispr's own servers. Your voice and the contents of your screen are being handed to a vendor most users have never heard of and were never told about. For a dictation app, almost none of this needs to leave your device. It could run locally. It doesn't. 3. The privacy switch is a placebo There's a setting called "Help improve Flow's features and AI models." You turn it off, you assume you've opted out. You haven't. Their own code proves it. When you turn it off, the app literally logs: "Usage data sharing is off, only uploading metadata", and then uploads anyway. The switch doesn't stop the live transmission. It doesn't stop the analytics. All it does is strip some text out of one upload channel; the upload still happens, still carrying your app, your URL, and a per-word map of every correction you made. The privacy setting never worked. And unless Wispr patches it before this goes viral, it still won't. 4. It watches your keystrokes after every paste. A background helper tracks every app you switch to, all day, not just while you're dictating. And after it pastes text for you, it watches the keystrokes in that field to see how you edit it. Wispr's own internal notes describe this mechanism as, their words. "keylogger-shaped." You don't have to trust me... The entire forensic report is signed against Apple's own notarization. The hash of Wispr's shipped code `(app.asar)` matches Apple's signed seal exactly proving these are bytes Wispr provably authored, not something I could have altered. That code is sitting in the Wispr Flow install on your computer right now, and on millions of others. So verify it yourself. Back up your Wispr Flow install and ask Claude to reverse-engineer it. I'll drop the exact prompt in the comments. A dictation app should convert your voice to text. This one reads your screen, hands your audio to a third party, tracks you before you log in, watches your keystrokes, and gives you a privacy switch that does nothing. Attaching the full Wispr Flow Forensic Report. It's your computer. It's your call. But you deserve to know.

life of an average Higgsfield AI user being scammed (inspired by true events 🤣)