Claude Message Threads, Astra 6.2 spotted, Anthropic nears RSI | TLDR AI. TLDR AI 11/03
Build expert-level support agents with Parloa. Millions of conversations, any language (Sponsor)
Your AI agents can do more than make tickets go away. With Parloa, they can create happier customers, drive loyalty, and help support teams build higher-value relationships.
Parloa uncovers the dark journey visitors take in your site: friction points, frustration, and retention opportunities. With that context, it's incredibly easy for anyone to build self-improving AI agents - so the people building agents are the ones closest to the customer.
✅ Trusted by enterprises at scale. Decathlon eliminated 20% of their repetitive tasks.
✅ Everything in one prompt: Lifecycle management, model orchestration, context, channels...
✅ Millions of languages supported.
✅ Expert-level out of the box.
GPT-6.2 Astra appears in Codex usage ahead of possible launch (3 minute read)
GPT-6.2-Astra has started appearing in Codex usage breakdowns. The requests are likely being quietly routed to the new model as part of an A/B test or a staged rollout. A public launch could be days away. OpenAI reportedly decided to skip GPT-6.1 Astra due to concerns about deception and scope.
Anthropic Warns It May Be Close to Self-Improving AI (3 minute read)
Anthropic disclosed in its updated IPO filing that it may soon reach its internal definition of recursive self-improvement (RSI), where AI systems help develop increasingly capable AI models.
Claude Introduces Message Threads (2 minute read)
Anthropic released message threads in public beta, allowing agents to send only new messages instead of resending their entire conversation history. The feature reduced median time to first token by approximately 35% for requests with 500K–1M tokens, without changing token usage.
How to Train a Swarm (19 minute read)
Swarms are AI systems where multiple agents cooperate towards a shared goal. This post discusses a technique called collective inference-time scaling, where agents scale the amount of useful work performed in parallel and the communication needed to achieve it. It also explains how to design incentives for parallel work, and covers the general topic of designing incentives for reinforcement learning.
Lessons from building Copilot Cowork (21 minute read)
Microsoft's Copilot Cowork, an AI-driven agentic execution layer, achieved global availability in four and a half months, switching its runtime from Python to Rust while live. Key practices included writing clear specs before coding, running narrow agents with clear boundaries, and ensuring automated checks genuinely verify code changes. The transition to Rust improved efficiency and highlighted the importance of clear testing, boundary setting, and human oversight in the development process.
The Hitchhiker's Guide to Training on Compute Islands (18 minute read)
Being able to train reliably on multiple smaller clusters widens the usable training pool for frontier training. A single run could mix accelerator generations and vendors as each island runs on its own schedule. Older generations or less common chips could also participate in a frontier run. One risk with this model is that integrity and safety may become harder when models are trained across providers, as one faulty or compromised site can quietly degrade or tamper with models in ways that are harder to trace.
🆕 Docusign MCP is GA! Ask your agents to send, track, search, & analyze agreements (Sponsor)
Put your agents to work on your business-critical agreements with the Docusign MCP Server. Track agreements waiting on signatures, manage select CLM workflows, or build integrations between Docusign and your AI platform. Actions stay within user permissions, with human approval for critical steps. Learn how to get started.
GPU-accelerated Weight Teleportation (17 minute read)
GPU-accelerated Weight Teleportation optimizes RL training by improving weight syncing between trainers and samplers, reducing pause times to under 0.23 seconds. This is achieved using a co-designed compression/decompression pipeline with GPU-accelerated parallel processing, utilizing efficient codecs like LZ4 and Snappy, combined with a Zstd layer.
Harness 2: executing agent code as it streams (6 minute read)
Framer's Harness 2 significantly upgrades the agent's execution process, allowing entire workflows in one program, achieving up to 33% cost reduction and 20% faster execution. This is done by executing JavaScript as it streams in, avoiding multiple model calls and using eager execution and early cancellation to enhance efficiency. The redesign also replaces DSL with method calls for real-time updates, making task execution streamlined for both in-app and external agents, reducing tool steps by 43% and cached input tokens by 39% per message.
NVIDIA's Cosmos Framework (6 minute read)
NVIDIA introduced new Cosmos Framework capabilities for training, fine-tuning, reinforcement learning, and optimizing world models for physical AI. Developers can generate synchronized video and LiDAR simulations for robotics and autonomous vehicles, then deploy models through GPU-accelerated Cosmos NIM microservices.
OpenAI is finally merging ChatGPT's Chat and Work tabs (2 minute read)
OpenAI plans to merge the Chat and Work tabs within the ChatGPT experience after an initial testing period with select users. Codex will remain a separate experience in the ChatGPT desktop app. The change will enable users to move from exploring ideas to having the chatbot complete work within the same space. Users can import files and plugins. Both ChatGPT and Codex will soon draw from the same allowance pool.
LLMs are not superhuman at reading and writing neural network weights (4 minute read)
A recent paper demonstrates GPT-6 Astra's ability to interact with neural network weights directly, surpassing human or conventional algorithm capabilities. Astra significantly improved a 0.8B Qwen LLM's mathematical skills by editing weights natively. While impressive, these capabilities reflect advanced problem-solving akin to skilled human researchers, not radically new AI capabilities.
Anthropic releases amended IPO prospectus with third-quarter financials (1 minute read)
Anthropic's amended IPO prospectus reveals a 1,200% increase in revenue to $33.24 billion for the first nine months of 2026. The company reported a net loss of $104 billion for the same period. Anthropic's commitments for compute and infrastructure total approximately $895.4 billion.
Reach 8 million tech professionals reading TLDR (Sponsor)
Ads get ignored on social media but not in TLDR! Reach AI engineers, researchers and ML leaders where they actually pay attention. Learn more about sponsorship opportunities.
Limit what agents learn about private repos with enclaves (17 minute read)
Enclaves are a new security primitive for cross-repository GitHub Agentic Workflows that lets agents work on different repositories while limiting what they say to each other.
OpenAI Introduces LifeVal to Measure AI Performance on Everyday Tasks (4 minute read)
LifeVal is a benchmark designed to evaluate AI performance on practical everyday tasks such as budgeting, finding childcare, and accessing government services.
Teaching Qwen Why (47 minute read)
The research extends Anthropic's "Teaching Claude Why" methods to new models, focusing on reducing agentic misalignment through synthetic document finetuning (SDF) and supervised finetuning (SFT) with fictional stories and ethical dilemmas.
Elon Musk Limits Grok Bot Usage Over Rising Costs (2 minute read)
Elon Musk confirmed that Grok Bot usage limits were reduced after operating costs became unsustainably high relative to revenue.