Google manages the "skills" used by AI agents just like regular products. A skill is a set of instructions that tells an AI what to do, so even a small change can alter how the AI behaves. For this reason, every change goes through automatic checks and evaluation before it is deployed.
Google has shared an inside look at how it built and maintained quality for "agent skills," which help AI coding tools work smarter and more accurately. It explained the quality control problems that came up as many teams got involved and how it solved them.
Meta, the parent company of Facebook, has acquired Conductor, a startup that builds software using multiple AI agents. The move is seen as part of Meta's strategy to strengthen its AI agents and coding capabilities.
Cat Wu announced that she joined OpenAI on November 4, 2026, to lead product management for its core products. She left Anthropic just one day earlier. Wu said she wants to build agents (AI programs that carry out tasks on their own) that users can hand work to while still staying in control.
Meta has acquired Conductor, a startup that builds software by running several AI agents at the same time. The deal amount was not disclosed. Conductor has stopped accepting new sign-ups and is gradually shutting down its service.
This video from the Cole Medin channel examines whether open-weight models (AI models whose weights are publicly released) can be a practical choice for agentic engineering (where AI carries out multi-step work on its own). According to the introduction, their benchmark scores are similar to top commercial models, but they fail to follow instructions and break down on long tasks.
Kakao CEO Jung Shin-ah pointed out the limits of global AI agents that directly operate web browsers. Kakao said it will pursue a strategy combining direct connections with major services and computer use agents (CUA, AI that views the computer screen and operates it directly).
We ran cua-driver, an open-source tool for computer control, on a MacBook connected to an AI agent. We ran five GUI scenarios directly. Several of them got stuck, and we found other routes to get results.
In Cursor, you can describe an app in words, and an AI agent will build it and deploy it. Finished apps can be shared with teammates through a single link. This feature is available on paid plans.
Conductor, a US startup that built a tool for running several AI coding agents at the same time, announced it is joining Meta. The purchase price was not disclosed. Conductor will be renamed Meta Conductor, and its existing service policies have been set out.
Meta has acquired Conductor, an AI coding startup whose tools let multiple AI agents help with software development. Meta confirmed the acquisition but did not disclose the deal terms.
NVIDIA has built a system in which multiple AI agents work together to help run its own AI data centers. The agents find problems, connect the causes, and show recommendations to the responsible engineer through chat. The goal is to reduce the effort of checking multiple dashboards one by one.
Cognition has released SWE-2 Flash, a coding model. According to the company, it is three times faster and one-fifth the price of its larger sibling, SWE-2. It targets everyday software work such as automation, code review, and feature development, and it is free to use until November 30, 2026.
HIX AI has released PPT Generation Agent 2.0. Instead of building presentations from a blank screen, it now offers more features that create slides from files and templates users already have. The company says it will let working professionals spend more time on planning.
Anthropic has added partner agents and products in security, developer infrastructure, and creative and productivity areas to its Claude Marketplace. More than 25 products are now listed. Eligible business customers can buy these products using part of their committed Anthropic usage budget.
Amazon AWS Machine Learning blog · Nov 5, 03:46 KST
AWS has shown how to connect decision models to Amazon Bedrock AgentCore Evaluations, a tool for testing AI agents. A decision model is a small specialized model that picks an answer from a fixed set of options and also reports how confident it is. The article covers how to connect two external decision models to the evaluation tool.
Conductor, a coding agent startup, has announced it is joining Meta. Its cloud service, Conductor Cloud, will shut down on November 20, 2026. The Mac app will be supported until early 2027 and then gradually discontinued.