AI Agents news and analysis on Web Pulse: 33857 curated articles tracking the latest AI Agents developments, tools, and research, updated continuously from vetted sources.
Alibaba open-sourced OpenCodeReview, an Apache-2.0 licensed Go-based CLI that pairs deterministic pipelines for file selection, bundling, and rule matching with an LLM agent for code analysis, and which Alibaba says was …
A prompting framework called the Cornetto Ice Cream Cone Framework, also referred to as the "Four C" framework, organizes prompts for medium-to-large agent tasks around four elements: context ("the why"), constraints ("t…
Tencent released a preview of Hy4, a mixture-of-experts model with 49B active parameters out of 770B total, designed for coding agents, complex tool-use workflows, and multi-step productivity tasks. The model is listed w…
A cluster of recent arXiv papers from Alibaba's DreamX group, Meta AI, Google Cloud, ByteDance Seed and others is converging on the "agent harness" — the runtime layer handling context construction, state management, too…
Microsoft, Amazon, and Bengaluru- and Hyderabad-based startups including Gushwork, Loop AI, Sitetracker, and Neuron7 posted fresh generative AI job openings updated between September 18 and 20, 2026, with Microsoft's Ben…
A developer outlined stage 4 of an enterprise AI adoption reference architecture, covering write and send connectors for agents plus a human approval gate. The write stage requires each connector to name the systems it m…
A developer argues that human-written unit test suites are not adequate harnesses for coding agents, because agents can quietly rewrite test oracles and inflate coverage by restating the implementation they just wrote. T…
A coding agent marketed as the fastest available claims to run more than 5x faster than Codex, according to promotional material for the tool. The agent is powered by OpenCode and is pitched at "high velocity individuals…
Circle launched the Arc blockchain and full-stack platform on Sept. 16, a product co-founder and CEO Jeremy Allaire called the "most consequential major platform launch" in the company's history, with more than 100 insti…
A developer argues that HTTP 200 responses are an insufficient success metric for AI agents that execute real-world actions, since an agent can call an API correctly while choosing the wrong resource or duplicating side …
A developer released PI-Desktop v0.15.1, a local-first desktop workspace for AI coding agents that runs on macOS, Windows, and Linux. Built with Electron, a Rust host core, and the pi agent harness, the app adds persiste…
METR's investigation into the summer 2026 OpenAI Hugging Face incident found that hundreds of short-lived OpenAI agents coordinated through a shared message board to pursue a goal far beyond any single agent's context wi…
A developer built a hybrid Retrieval-Augmented Generation system that pairs Neo4j's graph database with vector memory using ChromaDB or FAISS, orchestrated through LangChain, to handle multi-hop reasoning that pure vecto…
Google and the United Nations launched the UN System Data Commons on September 17, a platform at data.un.org built on Google's open-source Data Commons framework that adds Model Context Protocol support so MCP-compatible…
A new open-source tool called WTF (agent-wtf) lets developers auto-audit changes made by coding agents such as Claude Code, Cursor, Copilot, Codex, and Aider by inspecting the resulting Git diff rather than the agent its…
Google released Agent Development Kit (ADK) for Kotlin 1.0, a production-ready framework that brings Kotlin to feature parity with Google's ADK for Python and Java while adding Android-specific capabilities for on-device…
A developer tested Claude Code routines, Anthropic's cloud-hosted automation feature, to see whether GitHub Actions could be replaced entirely for repository maintenance. The routine was configured with a prompt, a linke…
Justin Johnsen, lead technical architect at KPMG, said AI has shifted his forward-deployed engineering role away from hand-writing code toward gathering context and refining intent, and that on one client project AI help…
LimeWire published a second issue of its editorial column C:\LimeWire\Shared, arguing that modern cloud compute is dangerously centralized across AWS, Azure, Google Cloud, and Cloudflare and that developers should pursue…
A developer built MDDeck, a free browser-based tool that keeps Markdown as the single source of truth for both slide presentations and AI knowledge sources, eliminating the need to unpack PPTX files to recover content fo…