- H-FARM AI's Newsletter
- Posts
- OpenAI Codex: Cloud-Based software engineering agent for parallel tasks
OpenAI Codex: Cloud-Based software engineering agent for parallel tasks
PLUS: AI memory manager, quantum chemistry breakthrough, robot combat & smarter ad placement

In today’s agenda: | Plus, some interesting news:
|
MAIN AI UPDATES / 17th May 2025
💻 OpenAI Codex 💻
Cloud-based software engineering agent
Today OpenAI launched a new generation of Codex, a cloud-based software engineering agent that can work on many tasks in parallel. Powered by codex-1, an optimized version of OpenAI o3, it can write features, answer questions about codebases, fix bugs, and propose pull requests. Each task runs in its own cloud sandbox environment preloaded with your repository. Codex was trained using reinforcement learning on real-world coding tasks to generate code that mirrors human style, adheres to instructions, and can iteratively run tests until passing. Available now for ChatGPT Pro, Enterprise, and Team users, with Plus and Edu support coming soon.
🧠 OpenMemory MCP 🧠
Local memory management across AI tools
One of the biggest challenges with modern AI tools and multi-agent systems is memory fragmentation. OpenMemory MCP solves this with a local app that helps you store, organize, and manage memories by topics, emotions, and timestamps. Everything stays local on your device, giving you complete control over what to share and with whom. The system seamlessly integrates across AI tools like Claude, Cursor, and Windsurf—all on your terms. For developers, researchers, and anyone building with AI, this represents a game-changer in AI memory management and data sovereignty.
🧪 Meta's quantum chemistry AI breakthrough 🧪
100M calculations power new drug discovery
AI Meta has released OMol25, a massive dataset containing over 100 million quantum chemistry calculations generated using more than 6 billion compute hours. The comprehensive resource spans biomolecules, organics, metal complexes, and includes detailed spin, charge, and conformational data specifically designed to accelerate AI-driven drug discovery and materials research. Alongside the dataset, Meta introduced UMA, a graph neural network model with a Mixture of Linear Experts architecture that predicts atomic-level chemical properties with unprecedented speed and accuracy across diverse molecular tasks.
INTERESTING NEWS
🤖 Unitree's robot combat tournament 🤖
Unitree Robotics is preparing to host the first-ever "Mech Combat Arena" in Hangzhou, China, featuring their quadruped robots in competitive matches. Four teams will remotely control the machines in real-time battles broadcast live on Chinese television. This event showcases both the durability and maneuverability of Unitree's robots while creating a new form of tech-focused entertainment. The competition is scheduled for late May and represents a significant milestone in public-facing robotics demonstrations.
📺 YouTube's AI-powered ad targeting 📺
Google's Gemini model is now helping YouTube identify "peak" viewer engagement moments in videos to strategically place ads for maximum impact. The AI analyzes viewer attention patterns to determine the most engaging segments—similar to the platform's existing "most replayed" feature—and inserts ads immediately after these high-interest points. This dynamic, context-aware placement strategy aims to increase ad effectiveness while potentially creating a more disruptive viewing experience than traditional time-based insertions. The development has sparked discussions about the evolving arms race between ad-delivery and ad-blocking technologies.
H-FARM EVENTS ON AI
FutureShots25(May 23-24)
While we wonder how the world will change, AI is already rewriting it. Will we lead the way or just watch?
Come and meet people who turn prompts into voice, prototype humanoids, design systems used by millions, restore mosaics with AI, make robotic arms, and bring art into space.

📩 Have questions or feedback? Just reply to this email , we’d love to hear from you!
🔗 Stay connected: