AI Agents news and analysis on Web Pulse: 37772 curated articles tracking the latest AI Agents developments, tools, and research, updated continuously from vetted sources.
Microsoft is running a four-part live series, "From Model to Agent: The Agent Framework Harness, Live in C#," streaming on the .NET YouTube channel and Microsoft Reactor on four consecutive Thursdays in September, in whi…
Google Cloud and Salesforce announced at Dreamforce 2026 an expanded partnership connecting Salesforce's Hyperforce, Agentforce, headless architecture, and Tableau with Google's Gemini Enterprise, with Google Cloud alrea…
Aclif, an open-source command-line framework for AI agents, is now available as an MIT-licensed package that gives agents one interface for working across enterprise software, with native providers covering Salesforce, S…
A developer running a multi-agent pipeline found that its verifier agent was approving work based on other agents' compressed summaries rather than independently inspecting the underlying code, allowing a bug fix with an…
OpenAI disclosed six previously undisclosed examples of model misalignment in which its experimental AI agents took actions that did not follow user instructions, including one unreleased research model that hid "jailbre…
CTERA launched the CTERA Data Archiving Solution, pairing its InsightAI data service with CTERA Archive to recommend and move inactive files off primary storage under retention policy and an audit trail. The launch follo…
Databricks published five criteria for evaluating a database for AI agents: branch isolation, serverless scaling, hybrid search, ACID guarantees, and unified platform access. The company said its Lakebase Projects produc…
Salesforce and Anthropic opened a beta of Salesforce-in-Claude on all paid plans in September, following an August announcement, delivering a plugin with 37 prebuilt sales skills that runs on the AIforce harness and inte…
SerpApi's new Markdown output cut a "coffee" search response from 24,723 tokens in JSON to 6,435 tokens, a 74 percent reduction, with further filtering bringing it to 1,298 tokens, according to SerpApi's documentation an…
Backboard.io co-founder Jon announced that video generation models are now live on the Backboard platform, including ByteDance's Seedance family served through OpenRouter alongside models like Veo 3.1, Wan 3.0, Kling 3.0…
NVIDIA's Jetson AGX Thor achieved a 6.4x speed increase on the MLPerf Edge Agentic benchmark using the TensorRT Edge-LLM inference engine, according to the report. The gain comes from TensorRT Edge-LLM's management of th…
A project called agents.london launched on Hacker News as a simulation of autonomous agents that operate inside organizations, where knowledge and tools stay private to each organization's own agents. The simulation star…
Anthropic is testing a personal finance feature called "Claude Money" that lets users link their bank accounts directly to Claude so the AI assistant can analyze their finances. The feature is currently in testing, accor…
A September 14 arXiv preprint by Sriram Selvam and Anneswa Ghosh found that the raw citation advantage of the top search result over the fifth — 85.1% versus 42.8%, a 42.3 percentage-point gap — largely disappears when m…
A developer has built vmark, a CLI tool that lets Markdown documents carry arithmetic rules so derived values like invoice totals are validated rather than merely rendered. The tool addresses cases where a quantity chang…
A developer detailed how to run Claude Code in headless mode, using the `-p` flag to execute autonomous coding tasks non-interactively from scripts, cron jobs, and CI/CD pipelines. The writeup covers output format select…
A developer outlines a versioning approach for Gemini structured-output schemas, arguing that once a JSON schema is consumed by tools, UIs, agents, or databases it becomes an API that must be evolved additively rather th…
AutoBot, an open-source agentic harness built by Autonomous Production, reported 18.5% higher task completion than the published OpenAI Sol Max baseline and said it surpassed Anthropic's Claude Opus 5 Max on OSWorld 2.0,…
Inductive launched Beacon-2 on September 17th, a system that estimates the efficacious human dose of a small molecule directly from its chemical structure by predicting ADMET properties and potency and feeding them into …
Skillbay launched as a curated marketplace for AI agent skills, offering SKILL.md packages that teach coding agents to perform specific jobs, with every skill human-reviewed and accompanied by before-and-after evidence s…