Agents and Copilots: AI News Week Ending 02/21/2025

“MAX 25.1 is a huge release: enabling new agentic workflows ( https://x.com/clattner_llvm/status/1892294758007263724

“Excited to introduce the Perplexity Deep Research Agent: available for free to all users. Paid users only need to pay $20/mo to access an expert level researcher on any topic for 500 daily queries, and need to wait less than three minutes for getting a full research report. https://x.com/AravSrinivas/status/1890464738951233536

“Github 👨‍🔧: Framework for building scalable agentic applications. → Open-source TypeScript library for building production-ready multi-agent systems. → Provides production-grade features including memory management, agent state serialization, and secure sandboxed code https://x.com/rohanpaul_ai/status/1890550582999445686

““We asked users if they would switch to an alternative Android digital assistant like Perplexity. We received nearly 6,000 votes in our site-wide survey, with over 52% of voters saying they would switch from Gemini to Perplexity without hesitation.” https://x.com/AravSrinivas/status/1891626705858576737

“you can now deploy AI agents to browse the internet, automate tasks, and execute complex workflows on your behalf. this is version zero of the tech (chatgpt operator etc). imagine version ten!!!!! how are you sleeping at night!!!” / X https://x.com/gregisenberg/status/1888240832479682610

“.@TallyApp is building AI agents for accounting firms. Tally agents operate in a firm’s existing systems, automating repetitive work across accounting, tax, and audit, helping teams be more efficient. https://x.com/ycombinator/status/1889751328836006052

“🚀 LangChain in Atlanta! 🚀 Join us next Thursday, February 27th for an evening of AI at the Honeywell offices with LangChain CEO, @hwchase17” / X https://x.com/LangChainAI/status/1892301688138047842

“We’ve just released the coolest feature ever in smolagents: you can now share agents to the Hub! 🥳🥳 And any agent pushed to Hub get a cool Space interface to directly chat with it. This was a real technical challenge: for instance, serializing tools to export them meant that https://x.com/AymericRoucher/status/1890431468700332366

FinRL-DeepSeek – new trading AI agents combining Reinforcement Learning with Large Language Models — AstroWind https://melwy.com/finrl_deepseek

“Ticket-to-PR on @linear 🚀 Rig up a SWE agent to respond to linear events + create PRs. Fully open source. Deploy your own. Here’s how 👇” / X https://x.com/mathemagic1an/status/1892691060192575746

“Introducing Proxy 1.0 – the world’s most capable web-browsing agent. https://x.com/convergence_ai_/status/1892129466610073931

“Here are some of the coolest things that I saw built with v0, lovable, bolt, & replit agent this week. 🧵 (save for later) https://x.com/AtomSilverman/status/1890135846642413696

“we just used @Replit agent to build a browser agent 🤯 you might not want to give @OpenAI your $200 just yet… this cost $6.14 to build. here it is browsing @ProductHunt and making a list from a prompt… https://x.com/MakerThrive/status/1888998516250304900

“Introducing TypeScript AI Agent Framework @PraisonAI 👇 Code below 🚀 JavaScript Multi-Agent AI Framework 🔌 Compatible: Node.js (ESM and Common JavaScript) 💻 Installation: npm install praisonai 🤝 Multi Agent Collaboration @AtomSilverman @Saboo_Shubham_ https://x.com/MervinPraison/status/1885218123101442080

“Super E Z these days to understand and build AI agents. Check out how I built one with Langflow and Composio! https://x.com/MelGoesTech/status/1889061741730107425

“MLGym creates an environment for SWE-agent to try and optimize various ML systems, including a literal language model. Cool!” / X https://x.com/OfirPress/status/1892927872085479813

“I built a multi-agent framework that pre-approves loan applications in under 5 minutes 🎉 1. Integrated 5 agents that make use of @Experian and @Plaid APIs for real-time credit and financial verification 2. The agents look up credit score, DTI, LTV, assets, property valuation https://x.com/n_sri_laasya/status/1890132203419627562

“LangGraph – powering agents and agent platforms from LinkedIn, Uber, Klarna, Replit, and more 👀 https://x.com/hwchase17/status/1892692277937557789

“Existing LLM agent frameworks are too complex, requiring programming expertise, limiting their accessibility to only a tiny fraction of the global population. This paper introduces MetaChain to solve this accessibility gap. MetaChain enables anyone to build and use LLM agents https://x.com/rohanpaul_ai/status/1891404810584572085

“Announcing Native Mobile App support on Replit. Now you can build iOS and Android apps that you can take all the way to the App Store without writing any code, powered by Replit Assistant. This is early access—full Agent support coming soon! https://x.com/amasad/status/1888727685825699874

“AI agents don’t just work perfectly out of the box. How do you know if they’re actually performing well? In our new short course, Evaluating AI Agents, built in collaboration with @ArizeAI, you’ll learn how to track, test, and iterate your agent’s performance—from individual https://x.com/DeepLearningAI/status/1892250339878818124

“The best revenue teams use @aomniapp to 10x their output. We’ve helped hundreds of companies in the past few months exponentially grow revenue with AI agents, and now we just raised a $4m seed! Check out Aomni in action in 🧵 https://x.com/dzhng/status/1891897453831491838

“📈 A multi-agent Financial Analyst with #crewai 💰 The user submits a query and then: ✅ Query Parser Agent extracts structured outputs form the query using #pydantic ✅ Code Writer Agent writes #python code to visualize the stock data using #pandas, #matplotlib, and https://x.com/crewAIInc/status/1889437990214357125

Introducing Our Agent Leaderboard on Hugging Face – Galileo AI https://www.galileo.ai/blog/agent-leaderboard

“AI agents, like rambunctious interns, will need ways to keep their humans (we’ll be busy doing other things!) in the loop for approvals, adjustments, and next steps. If you’re building new UX patterns or technical protocols in this area, I’d love to collaborate. DM or email me.” / X https://x.com/jeff_weinstein/status/1890269732538040493

“If you’re building AI Agents, this is the only podcast you need Here are my favourite insights from the latest episode of the AI Agents podcast with @kwindla and @KaranVaidya6, 1. Prioritize Low-Latency for Voice Interactions: If your AI agent uses voice, choose WebRTC over” / X https://x.com/_Prathit/status/1888612348140880145

“📽️ Just created an agent that: 1⃣ Searches @YouTube for recent popular videos that match a specified topic 2⃣Summarizes the videos 3⃣Drafts (or sends!) an email to me with the research summary. <2min to create using @ComposioHQ, Gemini 2.0, @YouTube, @GMail, and @llama_index. https://x.com/DynamicWebPaige/status/1887897486770974770

“✨ Kick off the AI Engineer Summit with us TONIGHT! ✨ LangChain and Tavily are hosting an evening meetup in NYC on February 19th. We’ll be diving into building reliable agents and the evolution of research agents. Join us for pizza, drinks, and great company!” / X https://x.com/LangChainAI/status/1892288795841876319

“Right now, I do not think an open Deep Research is possible – the magic seems to be in the o3 model more than the agentic use of search engines, etc The fact that the model “follows its curiosity” and delivers very low hallucination and insightful answers is an LLM quality thing” / X https://x.com/emollick/status/1889904648527683896

“NEW: Google introduces AI co-scientist. It’s a multi-agent AI system built with Gemini 2.0 to help accelerate scientific breakthroughs. 2025 is truly the year of multi-agents! Let’s break it down: https://x.com/omarsar0/status/1892223515660579219

“We are excited to launch an agentic search in @perplexity_ai! We’ve spent a lot of time optimizing its quality and speed to ensure it is available to all users and truly useful. Give it a try!” / X https://x.com/denisyarats/status/1890454335374434651

OpenAI’s Operator agent helped me move, but I had to help it, too | TechCrunch https://techcrunch.com/2025/02/04/openais-operator-agent-helped-me-move-but-i-had-to-help-it-too/

“ACU (Awesome Agents for Computer Use) This is a nice curated knowledge base of AI agents that can operate computers. It compiles research papers, projects, frameworks, and tools for “agents” that autonomously perform tasks via clicks, keystrokes, command-line calls, and API https://x.com/omarsar0/status/1887911960551043166

“End to end agent traces are still our bread and butter, but folks were asking for a more granular view of LLM calls. Especially w.r.t. model latency Putting the finishing touches on the new LLM logs viewer after an evening of yap engineering w/ Cursor (coming soon to prod) https://x.com/AlexReibman/status/1889945681944781049

“Is @fewsats and @Replit is all you need? This weekend we participated in the @coinbase hackathon in SF @kleffew94 We made the replit agent purchase the domain that it will use to deploy an app ITSELF! https://x.com/positiveblue2/status/1886214580726321422

“DeepSeek R1 was just the start—this new Chinese research from @Kimi_Moonshot lets RAG AI agents devour entire codebases and documentation with no context limits. Mixture of Experts and Sparse attention make near-infinite context possible. 🧵1/n 📌 Challenge of Long-Context https://x.com/rohanpaul_ai/status/1892535262879617101

“Never before has there been such an accessible knowledge work agent freely available, as Perplexity Deep Research. What are some other demos you would like to see with it? Please comment here. Also, what are the pain points you’re seeing today? We will look into supporting a” / X https://x.com/AravSrinivas/status/1891344192707969134

“Microsoft silently updated OmniParser on the hub 👀 60% faster than v1 – sub-second latency on a 4090! “OmniParser is a general screen parsing tool, which interprets/converts UI screenshot to structured format, to improve existing LLM based UI agent.” Bonus: you can try it https://x.com/reach_vb/status/1891467489030082875

“Introducing Taskade AI Kits! Bundle pre-trained AI Agents, Automations & Projects into a single link for one-click deployment. 🚀 Instantly share ready-to-use AI workflows 👥 Onboard teams & clients—no setup required ⚡ Deploy fully connected workspaces in seconds 🧵(1/) https://x.com/Taskade/status/1889323813680337136

“The Multimodal AI Agents Hackathon was a huge success! 🚀 200+ innovators & 40 teams lit up our HQ with energy, creativity, and epic iterations over the weekend. After 16+ hours of coding and pitching, here are the top 5 winning teams from the weekend and what they built! 🧵 https://x.com/weights_biases/status/1892254937901396068

“🔥 Just finished Unit 1 of the @HuggingFace Agents Course and wow – this thing is incredible! Crystal clear deep dive into AI Agents 🤯 Can’t recommend enough! #AI #HuggingFace https://huggingface.co/learn/agents-course/en/unit1/introduction

“Framework for managing AI agents https://x.com/tom_doerr/status/1890377316942967195

“I built this AI academic paper aggregator that automatically collects, summarizes, and shares academic papers. Features include daily paper updates and social sharing via @X using the @Replit agent with a simple prompt. Thanks, @amasad , give it a try https://x.com/MarcJ2327/status/1865615984473633097

“Multi-agent automation framework https://x.com/tom_doerr/status/1890127095289852293

“New short course: Evaluating AI Agents! Evals are important for driving AI system improvements, and in this course you’ll learn to systematically assess and improve an AI agent’s performance. This is built in partnership with @arizeai and taught by @JohnGilhuly, Head of Developer https://x.com/AndrewYNg/status/1892258190546653392

“OmniParser, groundbreaking screenshot parser for web automation, got even better and faster 🔥 It’s open-source (MIT), and you can plug this in model/agent of your choice: Qwen2.5VL, DeepSeek R1, 4o/o1/o3 mini Model and demo out on @huggingface https://x.com/mervenoyann/status/1891524621435830700

“I love this idea. Harshit made a UX for our AI hedge fund Under the hood: • @gumloop_ai for drag and drop UI • @LangChainAI for agent orchestration My dream for the AI hedge fund is to have an agent UX that runs daily. Very similar to this. If you are building an app layer” / X https://x.com/virattt/status/1890170921379414017

“OpenAI introduced “deep research,” an AI-powered agent that searches the web and generates detailed research reports. Currently available exclusively to ChatGPT Pro users, the deep research tool uses OpenAI’s o3 model to process information, ask clarifying questions, and” / X https://x.com/DeepLearningAI/status/1890476096409194976

“a taste of what AI agent interaction will be like” / X https://x.com/gdb/status/1891047216611037229

“I can now run models as good as the original ChatGPT locally on my iPhone Among other things it shows: how quickly AI is improving, the evolution of open weights models (this is Llama 3.1 8B), the opportunity for intelligence everywhere & the weakness of Siri AI compared to SoTA https://x.com/emollick/status/1891606625883955706

“I just open-sourced this web-testing AI Agent. It takes Gherkin scenarios as input and performs web testing on its own. PS: I used TypeScript 🤫 Source code: https://x.com/kevinpiac/status/1888544976939712533

“Can we use AI agents to automate the process of optimizing low-level GPU computations? The AI CUDA Engineer is a new @SakanaAILabs’s system that automates CUDA kernel optimization, improving how models run on GPUs. Benefits: – It can generate CUDA programs that outperform https://x.com/TheTuringPost/status/1892725955388702788

“I work on an AI agent called Bluewind. It uses success stories to generate new leads. Here is the one question I ask myself every hour of every day I work on it 🧵 (save for later)” / X https://x.com/wayne_hamadi/status/1890467482445513093

“🦜⚛️ useStream(agent) Integrate LangGraph into your React application with a single hook. Thread and state management, messages streaming and branching built-in. https://x.com/LangChainAI/status/1893023232938193168

“Microsoft just dropped OmniParser V2, looks incredible Turning Any LLM into a Computer Use Agent https://x.com/_akhaliq/status/1890546832784208080

“This week brought some crazy releases for AI Developers 🧵 • Conversation prediction AI • Build mobile apps w/ @boltdotnew • Open Source ChatGPT Search • Reasoning Augmented Generation by @superagent_ai • Open protocol for AI agents Here’s a roundup of everything 👇 https://x.com/ericciarla/status/1890499862749409700

“🔍 How LinkedIn enhances Sales Navigator with LangChain To refine LLM-powered features like AccountIQ for sellers, @LinkedIn built collaborative prompt engineering playgrounds using Jupyter Notebooks & LangChain. In this blog post, see how they: 🔹 Automated company research” / X https://x.com/LangChainAI/status/1890531416800383074

“I added stock prices to our AI financial agent Agent now shows you: • stock price • market cap • volume • historical prices (chart) No signup required to use the agent. All of the code is open source. https://x.com/virattt/status/1889458443515265066

“Magma A Foundation Model for Multimodal AI Agents https://x.com/_akhaliq/status/1892074819795005518

“Who doesn’t love building with AI on the weekend? 😎 Some automation, some innovation, and just enough time to wonder if you are training the AI or the AI is training you. 😂 Today, I built an MVP using @Replit AI agent. It is a gaming webapp that connects to a crypto wallet https://x.com/crypto_XZ/status/1888658127119347768

“🐘 LangMem SDK: long-term memory for AI agents Different types of memory enable different agent capabilities. We’re releasing an open-source library that implements some of these patterns so you can build agents that learn from interactions. The LangMem SDK makes it easier to: https://x.com/LangChainAI/status/1891881053897490772

“Our paper advocating that Fully Autonomous AI Agents Should Not Be Developed has garnered a lot of interest! https://x.com/mmitchell_ai/status/1891534146213666870

“Last week, @NWischoff asked for an “AI agent to underwrite a mortgage or even do a partially underwritten pre-approval in less than 24 hours.” @AgentOpsAI build a fully functional agent that pre-approves loan applications in under 5 minutes. Comment below if you are interested https://x.com/AtomSilverman/status/1891965335169008133

“🤖 AI Agent Memory Tutorial Build smarter AI agents with PostgreSQL-powered persistent memory using LangChain React agents. This tutorial shows how to implement database storage that enables your AI to maintain long-term conversations. 📺 Learn more: https://x.com/LangChainAI/status/1887909186899505174

“🤖💾 MongoDB x LangGraph.js Webinar Learn to build JavaScript AI agents using MongoDB’s data platform with LangGraph.js. This technical session covers Node.js integration and state persistence for powerful context-aware AI apps. Join us 2/25 11AM ET! 📅 https://x.com/LangChainAI/status/1891505383874625618

“Microsoft presents: Magma: A Foundation Model for Multimodal AI Agents – SotA on UI navigation and robotic manipulation tasks – Pretrained on a large dataset annotated with Set-of-Mark (SoM) for action grounding and Trace-of-Mark (ToM) for action planning. https://x.com/arankomatsuzaki/status/1892059107479224384

“🛠️ Faster prompt iteration with LangGraph Studio’s Playground integration Within LangGraph Studio, you can now view LLM calls directly in the Playground and iterate on prompts without re-running your whole graph. Speed up agent development by configuring system prompts without https://x.com/LangChainAI/status/1892288249441505444

HP $116 Million Deal for Humane Includes IP But No Ai Pin Device – Bloomberg https://www.bloomberg.com/news/articles/2025-02-18/hp-116-million-deal-for-humane-includes-ip-but-no-ai-pin-device

Humane’s AI Pin is dead, as HP buys startup’s assets for $116M | TechCrunch https://techcrunch.com/2025/02/18/humanes-ai-pin-is-dead-as-hp-buys-startups-assets-for-116m/

How We Ran A Project Without Any Meetings at Atlassian https://www.atlassian.com/webinars/business/how-we-ran-a-project-without-any-meetings-at-atlassian

“Deep Research as a small business incorporation legal consultant. Usually charged hundreds or even thousands of dollars an hour to offer this. Now free, only on Perplexity. https://x.com/AravSrinivas/status/1891563239240069245

“.@harvey__ai is THE vanguard AI app startup 🆕 @NoPriorsPod w/ Cofounder-CEO @winstonweinberg on: *capability improvement *co-pilots, end to end task completion, and product strategy for AI apps *selling AI to enterprise *hiring philosophy *what lawyers do in 5 years links 👇 https://x.com/saranormous/status/1890437612327874751

“Can LLMs earn $1 Million from software engineering? SWE-Lancer is a new Benchmark from @OpenAI that test if LLMs can do freelance tasks from Upwork valued at $1 million USD total in realworld payouts. SWE-Lancer includes independent engineering tasks (IC) from $50 bug fixes to https://x.com/_philschmid/status/1891780812497887289

“Today we’re launching SWE-Lancer—a new, more realistic benchmark to evaluate the coding performance of AI models. SWE-Lancer includes over 1,400 freelance software engineering tasks from Upwork, valued at $1 million USD total in real-world payouts. https://x.com/OpenAI/status/1891911123517018521

[2502.12115] SWE-Lancer: Can Frontier LLMs Earn $1 Million from Real-World Freelance Software Engineering? https://arxiv.org/abs/2502.12115

“OpenAI announces SWE-Lancer Can Frontier LLMs Earn $1 Million from Real-World Freelance Software Engineering? https://x.com/_akhaliq/status/1891721712296747126

Built this app in just 30 minutes on my phone using @Replit. The future is looking bright! 👀 https://x.com/inkko44/status/1888357988831834264

⚡️ Build an app and deploy it with Bolt.new and Expo – YouTube https://www.youtube.com/watch?v=iCwxkm2PkQE

Operator is now rolling out to Pro users in Australia, Brazil, Canada, India, Japan, Singapore, South Korea, the UK, and most places ChatGPT is available. Still working on making Operator available in the EU, Switzerland, Norway, Liechtenstein & Iceland—we’ll keep you updated! / X https://x.com/OpenAI/status/1892832374997631250

“so. this will never* happen because law firms have to be owned 100% by attorneys. so it’s impossible* to make a law firm that is also a tech company that rewards non-lawyers with equity” / X https://x.com/andersonbcdefg/status/1891329620558983633

“I built a FREE AI tool that creates ENTIRE websites, apps, and games in ONE CLICK I literally have my own digital developer that codes without me touching anything In this video I’ll show you how to set it up using Bolt DIY and Google Gemini 2.0 No coding skills required https://x.com/JulianGoldieSEO/status/1888977671838916784

Accelerating scientific breakthroughs with an AI co-scientist https://research.google/blog/accelerating-scientific-breakthroughs-with-an-ai-co-scientist/

“Google’s new “co-scientist” solved a complex microbiology problem in 48 hours, a task that took researchers at Imperial College London a decade to complete. Professor José R. Penadés tested the AI with a hypothesis about superbug resistance, and it correctly identified the https://x.com/rohanpaul_ai/status/1892746665225826321

“A no-code mode to start building AI apps just landed Gradio Sketch 🖌️ Install Gradio and just type: 𝚐𝚛𝚊𝚍𝚒𝚘 𝚜𝚔𝚎𝚝𝚌𝚑 in your terminal. https://x.com/_akhaliq/status/1892714927401337266

“Logic-RL Unleashing LLM Reasoning with Rule-Based Reinforcement Learning https://x.com/_akhaliq/status/1892776155674984509

“S* Test Time Scaling for Code Generation https://x.com/_akhaliq/status/1892772653947093190

“The AI-native (edge and LLM) proxy for agents. Move faster by letting Arch handle all the pesky heavy lifting in securing, processing, routing, and tracing prompts. Built by the contributors of Envoy. Key features include: 🛡️ Guardrails at the edge: reject jailbreak attempts https://x.com/_akhaliq/status/1891289618978316749

“most ai apps today are built with gradio 🔥” / X https://x.com/_akhaliq/status/1892260638191235244

“@Entelligence ❤️’s OSS! You can now understand any codebase with realtime documentation, tutorials, codebase chat, code reviews and more by simply changing https://x.com/Aiswarya_Sankar/status/1890502637360992451

“More apps should natively offer this. “Export for prompt” button” / X https://x.com/karpathy/status/1890208670732124372

Caidera | Home https://www.caidera.ai/

“I’ve collected 4,650+ website components on @landingfolio over 10 years. Here’s how I use them to create landing pages with @v0 https://x.com/dannypostmaa/status/1887736315980226742

“@lmstudio Speculative decoding in LM Studio! Can’t wait to play with it! https://x.com/cognitivecompai/status/1892252216515293381

FastTrackr AI – For Financial Advisors https://fasttrackr.ai/

Fiverr Go | AI-Powered Tools to Amplify Human Talent https://www.fiverr.com/go

“🛠️ I built my portfolio website using @lovable_dev https://x.com/gusgarza_/status/1888284667339575375

“Built a fully working email marketing tool using Replit in less than 30 minutes. Damn. https://x.com/harshdeeprapal/status/1888109370027958304

Introducing Radar — real-time protection against bots, fraud, abuse — WorkOS https://workos.com/blog/radar

“Generalist SWE agents like @DevinAI are useful for making dev discussions in Slack concrete. You can use it to spin up a sketch of an implementation and get others’ feedback, etc. It doesn’t necessarily need to take it to the finish line (merge a PR) to add value. https://x.com/mathemagic1an/status/1891712313599623461

“LLMs will solve document recognition. They will also solve entire document workflows that traditionally requires an entire backoffice team and a week’s worth of effort to solve. 1. Contract Review 2. Invoice Processing and Reconciliation 3. Compliance Reporting LlamaCloud is https://x.com/jerryjliu0/status/1890559184372134006

“One of the first use cases people discovered with LLMs was marketing copy generation (e.g. Jasper/Copy), but the way it started was very much one-shot content generation. But the world is going towards marketing agents as sophisticated multi-step, hierarchical systems that also https://x.com/jerryjliu0/status/1890220015296938464

“I built an internal PM tool in about 5 hours with @lovable_dev and @supabase , I might actually finish this one out and start using it for my own agency. https://x.com/imjesseshow/status/1887944280729120934

“Introducing jina-deepsearch-v1, it search, read, reason, search, read, reason, search, read, reason, search, read, reason, search, read, reason, search, read, reason, search, read, reason, search, read, reason, search, read, reason, search, read, reason, … 🔄 until the https://x.com/JinaAI_/status/1890410008590086278

“Built a simple web app using @v0 to track time since any date — displayed in years, months, weeks, days, hours, minutes, seconds, milliseconds, microseconds, nanoseconds, and picoseconds. https://x.com/juraj_stefanic/status/1888722445982437810

“just built an app faster than my mom could find a competitor on the app store using @lovable_dev for the first time😍 https://x.com/kathrynjc7/status/1889784795451842912

“Introducing Visual Edits You can now visually edit any styles in Lovable, for faster and more precise edits https://x.com/lovable_dev/status/1890075209748902304

“Built a simple Tetris like game in 5 minutes from bed using @lovable_dev My mission is to create our favorite games and publish them in the app stores for free without ads… too many freemiums and ad-based games out there. https://x.com/nomadicmarc/status/1888148679317446722

“Agents are cool and the tech behind Operator and Computer Use is super impressive. But I’m not convinced there are tons of super obvious consumer use cases… yet. That’s why everyone always cites the same obvious examples (“imagine you wanted to plan the perfect trip to” / X https://x.com/mignano/status/1889492610529763819

“Built in less than a day using: – @nextjs & @shadcn UI – @ClerkDev auth – @supabase db & @DrizzleORM – @cursor_ai & @v0 – deployed to @vercel Happy to share the codebase, but still a few bugs to fix.” / X https://x.com/MisbahSy/status/1889444112077643865

“Kraftful is great for summarizing user feedback. I also think very highly of their founder. 🙂” / X https://x.com/npew/status/1892970660508426742

“How I Built Good Supplements for my client in 2 Weeks Using @lovable_dev + @cursor_ai Good Supplements is a platform for biohackers and wellness enthusiasts to discover, organize, and share supplement stacks. Here’s how I built this entire MVP step by step. https://x.com/PrajwalTomar_/status/1888594521522376943

“Sakana AI’s new system, The AI CUDA Engineer, automates the creation of efficient CUDA kernels for machine learning operations. It translates PyTorch code into specialized GPU kernels, then refines them through evolutionary strategies to reach significant speedups. CUDA kernels https://x.com/rohanpaul_ai/status/1892717069092979059

“Okay @lovable_dev is awesome! Made this Website using only 7 prompts! 🔥 – Modify the UI/UX a bit – Added Authentication and Backend. Building More and More Websites using this.💪 Also used first time OBS recorder…will love your feedback!🙏 https://x.com/ShantanuK86/status/1889009541465149789

“> every line shared becomes collective momentum that accelerates the journey. > No ivory towers – just pure garage-energy and community-driven innovation. You can just tell, R1 “hands” wrote this. As I’ve been saying: a uniquely cyborgist lab. https://x.com/teortaxesTex/status/1892793229092827619

“Automate online tasks with agents https://x.com/tom_doerr/status/1889179182761943219

“Our new models for studying scaling laws are out! The Gemstones are 4K checkpoints (22 models) trained on 10T token combined, with varying architectures and learning rates. Here’s my fav new scaling experiment. It explains why industry has abandoned big dense models 🧵 (1/4)” / X https://x.com/tomgoldsteincs/status/1890426270372004085

“Built a habit tracker app with https://t.co/AL1Gg7ierH using an AI feedback loop And I documented the entire process step-by-step (17 pages) Comment “BOLT” and I’ll send you the guide (must be following so I can DM) https://x.com/TradeForOpp/status/1887922871713816734

“AI that learns from gameplay and generates visuals. This will be a new chapter in Gaming for the games you love! 🎮” / X https://x.com/yusuf_i_mehdi/status/1892244909484077354

“everyone who bought the $700 AI pin got literally rugged” / X https://x.com/nearcyan/status/1892049572546904319

“OpenAI Deep Research is not built for prediction (people always ask about stock & demand forecasts) it is built for analysis: make an argument about a point of view and the evidence to support it. This is what lawyers, accountants, academics, analysts, and entrepreneurs do a lot.” / X https://x.com/emollick/status/1890063434022347257

“A new version of @OpenAI’s ChatGPT-4o is now live on Arena leaderboard! Currently tied for #1 in categories: 💠Overall 💠Creative Writing 💠Coding 💠Instruction Following 💠Longer Query 💠Multi-Turn This is a jump from #5 since the November update. Math continues to be an area https://x.com/lmarena_ai/status/1890477460380348916

“If reasoning models had not been invented at the end of 2023 / early 2024, the AI hype would have died this year after the GPT-5 release. What they call GPT-4.5, really is GPT-5. But it would have been a disappointing one without the o3 reasoning on top. The promise of scaling” / X https://x.com/scaling01/status/1892733059000148137

“Here’s the final results – built in 15 mnts: Rundown: – I used @CodeGuidedev Starter kit with @boltdotnew – generated PRD + app flow with CodeGuide – uploaded docs to Bolt – used 21st .dev for Ready-made UI components – OpenAI API (for report gen.) – Jina ai (for scraping) https://x.com/cj_zZZz/status/1888627087650795614

“Perplexity just announced Deep Research (PDR)! I’m now testing and comparing it with OpenAI’s Deep Research (ODR). I still think the o3 variant powering ODR is a massive advantage. 20.5% (PDR) vs. 26.6% (ODR) on Humanity’s Last Exam. https://x.com/omarsar0/status/1890525249977872640

“Perplexity Deep Research is quite close to OpenAI o3 on the Humanity Last Exam Benchmark despite being an order of magnitude faster and cheaper. This is possible because DeepSeek is open source and cheap and fast. https://x.com/AravSrinivas/status/1890486069361025040

“This is my o1-pro workflow guide for building apps from a starting template. It’s honestly *crazy* good. 4hr+ detailed 6-prompt workflow to maximize your ability to build with AI. Great for beginners & experienced devs. Full tutorial here tomorrow at 12pm PT. https://x.com/mckaywrigley/status/1890457528087044401

“Introducing o3-mini Deal Finder 🛒 Have AI search the web and find the best place to buy a product. Simply enter the product name, and it will search the web providing a review summary and the best deal. Powered by @firecrawl_dev and OpenAI’s o3-mini. https://x.com/ericciarla/status/1892257110047805660

“Just a reminder that AI is so challenging to figure out because it is genuinely capable of doing PhD-level work in some areas while messing up basic tasks in closely related areas. And the abilities of AI are growing but unevenly. This jagged frontier explains a lot of confusion” / X https://x.com/emollick/status/1892215455554613386

“3/ AgentStack from @AgentOpsAI v0.3.4 – CLI command to create custom tools for any framework – CLI command to undo AgentStack actions – improvements to @firecrawl_dev – vision tools with @AnthropicAI – SQL tool added to core tools @braelyn_ai https://x.com/AtomSilverman/status/1890534557817974900

“🚀 CrewAI v0.102.0 is here! 🚀 Major improvement on CrewAI Flows Custom embedding support Improved LLM structured outputs JSON logging Multi-tab Excel knowledge sources Memory reset, and more—plus tons of bug fixes and doc upgrades.” / X https://x.com/joaomdmoura/status/1890068416033427855

“Optimizing Model Selection for Compound AI Systems Building with multiple LLMs to solve complex tasks is becoming more common. In a compound system, which LLM do you select for each call? Researchers from Microsoft Research and collaborators introduce LLMSelector, a framework https://x.com/omarsar0/status/1892945381174210933