{"date":"2026-10-12","title":"What happened in AI — Oct 12, 2026","generated_at":"2026-10-12T21:23:03Z","intro":["Production agent practice led the day: Anthropic, Coinbase and Spotify each described how they coordinate, evaluate or control agent-driven work at scale.","On the infrastructure side, vLLM reported up to 3.3x lower TTFT for multi-turn agent workloads, while Chinese labs drew funding and new model releases."],"highlights":["Anthropic explains how dynamic workflows coordinate Claude agents on larger projects.","Coinbase details building, evaluating and post-training a fraud agent.","Spotify more than doubled monthly code changes with AI and is tightening quality controls.","vLLM bi-directional KV transfer cuts multi-turn TTFT by up to 3.3x.","DeepSeek backs a $400M robotics lab; Z.ai ships GLM-Legal."],"article_count":14,"categories":[{"name":"Agent Orchestration and Production Agents","slug":"agent-orchestration-production-agents","summary":"Teams are publishing how they coordinate, evaluate and post-train agents in production, not just prototype them.","articles":[{"title":"How Anthropic’s Dynamic Workflows Coordinate Claude Agents","summary":"Anthropic's AI Engineer New York talk applies Unix and Linux design lessons to coordinating Claude agents on larger projects without overwhelming developers.","source":"latent_space","url":"https://www.latent.space/p/claude-dynamic-workflows","published":"2026-10-12T18:18:17Z"},{"title":"Coinbase Built, Evaluated, and Post-Trained a Fraud Agent","summary":"Coinbase describes building, evaluating and post-training a fraud-detection agent, a rare end-to-end production case study.","source":"hackernews_ai","url":"https://www.coinbase.com/en-gb/blog/part-one-how-coinbase-built-evaluated-and-post-trained-a-fraud-agent","published":"2026-10-12T19:39:46Z"},{"title":"Spotify More Than Doubled Monthly Code Changes as AI Accelerated Development","summary":"Spotify says AI-assisted development more than doubled monthly merged changes, and it is revising testing and reliability controls in response.","source":"infoq_ai_ml","url":"https://www.infoq.com/news/2026/10/spotify-ai-quality-controls/?utm_campaign=infoq_content&utm_source=infoq&utm_medium=feed&utm_term=AI%2C+ML+%26+Data+Engineering","published":"2026-10-12T13:57:00Z"},{"title":"GitHub Copilot app for Beginners: How to teach Copilot about your codebase","summary":"GitHub's walkthrough covers repository custom instructions, skills, custom agents and MCP servers such as Playwright.","source":"github_blog_ai_ml","url":"https://github.blog/ai-and-ml/github-copilot/github-copilot-app-for-beginners-how-to-teach-copilot-about-your-codebase/","published":"2026-10-12T17:43:59Z"}]},{"name":"Inference Serving and Endpoint Operations","slug":"inference-serving-endpoint-operations","summary":"Serving work targets multi-turn agent latency and safer endpoint rollouts.","articles":[{"title":"Improve TTFT for Multi-Turn Agentic Workloads with Bi-Directional KV Transfer in vLLM","summary":"vLLM's bi-directional KV transfer lets prefill nodes reuse KV computed on decode nodes, cutting multi-turn time-to-first-token by up to 3.3x.","source":"vllm_blog","url":"https://vllm.ai/blog/2026-10-12-bidirectional-kvxfer-multiturn-agentic-workload","published":"2026-10-12T00:00:00Z"},{"title":"Endpoint Candidates: evaluate and promote inference endpoints","summary":"Modal's Endpoint Candidates tests inference endpoints on quality, latency, throughput and cost before promoting them.","source":"modal_blog","url":"https://modal.com/blog/endpoint-candidates","published":"2026-10-12T00:00:00Z"},{"title":"Deploy decision models with llama.cpp on Amazon SageMaker AI","summary":"AWS shows llama.cpp on SageMaker AI serving OpenAI-compatible chat completions via its Deep Learning Container.","source":"aws_ml_blog","url":"https://aws.amazon.com/blogs/machine-learning/deploy-decision-models-with-llama-cpp-on-amazon-sagemaker-ai/","published":"2026-10-12T15:51:46Z"},{"title":"Launch HN: Agent37 (YC F26) – Low-cost agent sandboxes for running agents","summary":"Agent37 pitches agent sandboxes at under 10% of typical provider cost, citing an average of $0.37 per sandbox per month.","source":"hackernews_ai","url":"https://news.ycombinator.com/item?id=50059096","published":"2026-10-12T16:46:34Z"}]},{"name":"Chinese Open-Weight Labs: Models and Capital","slug":"chinese-open-weight-labs-models-capital","summary":"DeepSeek funding, Z.ai's new model and a Qwen-based Microsoft model show where Chinese labs are moving.","articles":[{"title":"Z.ai Unveils GLM-Legal, an Agentic Model for Legal Workflows","summary":"Z.ai released GLM-Legal, an agentic model aimed at legal workflows.","source":"search_cn_open_weight_labs","publisher_name":"Unite.AI","publisher_domain":"unite.ai","url":"https://news.google.com/rss/articles/CBMiiAFBVV95cUxPNmtnUTl5TGZyXy02MlJfZDc0c2t2ZXdpYnVyQWtReEJpWHV2Qk5qUXpLaG9fakZwNGtfeTFkYkx3QzcxZFlTajBjTkEtOWFqaVUxZzVLSWVMeXB0UkFSUy1LdjczLUh6UFY3ZS1SekNnb1pnLVBOaURHbVQzak14X2RYYzYtZy1R?oc=5","published":"2026-10-12T16:37:10Z"},{"title":"Microsoft-Decision-1 runs on Alibaba’s Qwen and costs $0.042 a million","summary":"Microsoft-Decision-1 runs on Alibaba's Qwen and is priced at $0.042 per million tokens, per The Next Web.","source":"search_cn_open_weight_labs","publisher_name":"The Next Web","publisher_domain":"thenextweb.com","url":"https://news.google.com/rss/articles/CBMid0FVX3lxTE5ELTNqd3JiMVpPRHN3Y0ZQYVFMUUtFREVMRjlkZVVqT01DVzVZZmg1Zjh0SjAtZFVyeXVpNlhXNHpaUE04RzUzTEE1cVVIX1FCWnFrX0EwbHg5aVBjY29kSmhaSndUWEYzVHZXdzRDOVdxMU9qZEhF?oc=5","published":"2026-10-12T09:01:53Z"},{"title":"DeepSeek Backs Ex-Tencent AI Scientist’s $400 Million Robot Lab","summary":"DeepSeek is backing a roughly $400 million robot lab from an ex-Tencent AI scientist (Winverse AI), per Bloomberg.","source":"search_cn_open_weight_labs","publisher_name":"Bloomberg.com","publisher_domain":"bloomberg.com","url":"https://news.google.com/rss/articles/CBMiswFBVV95cUxPMjE5N1J0MDF2LWVTMDR2UUJGdXVrU0lNdVRBSzRuNU5UbXJNdTh0UmVCQXZtV1JYQi01TnFNS0pLNzVxU1BIQ25OOWtUdjZvR0JuQVU3a1lfTGdMLW1rTGg0bWc4YWRnZGUyenBlSXhZTmM0UmJzUGZNcC1lWTNBZGJmZ3NtNHBEQkNsUVVNYS1ZaGszaGtmUmZqa3VYVHVyWDBVMFJLcnBxX1AwTnBwa3FnUQ?oc=5","published":"2026-10-12T11:42:26Z"}]},{"name":"Developer Tooling and Evaluation","slug":"developer-tooling-evaluation","summary":"New open tools and benchmarks target agent use and agent measurement.","articles":[{"title":"Hello Python 3.15, goodbye 3.10, and now I can use 3.11","summary":"Python 3.15 is out; Simon Willison points to the What's new guide and notes dropping Python 3.10 support.","source":"simon_willison","url":"https://simonwillison.net/2026/Oct/12/python-3-15/","published":"2026-10-12T18:15:38Z"},{"title":"A YouTube Transcript CLI Built for Humans and AI Agents","summary":"An open-source CLI fetches YouTube transcripts for both humans and AI agents.","source":"hackernews_ai","url":"https://github.com/nmdra/yt-transcript","published":"2026-10-12T16:59:11Z"},{"title":"OpenStocks, a public earnings forecasting benchmark for AI agents","summary":"OpenStocks is a public earnings-forecasting benchmark for AI agents.","source":"hackernews_ai","url":"https://openstocks.com/","published":"2026-10-12T12:28:05Z"}]}]}