AI Agents news and analysis on Web Pulse: 13376 curated articles tracking the latest AI Agents developments, tools, and research, updated continuously from vetted sources.
Weill Cornell Medicine researchers published EmulatRx in Nature Communications on July 7, 2026, a five-agent LLM framework for clinical-trial design using real-world patient data from EHR datasets such as MIMIC-IV and IN…
A developer argues that focusing on token optimization for AI agents is a distraction from the real cost drivers, which are agentic loops involving multiple subagents and long-horizon execution. The developer introduces …
A developer evaluates six LLM observability tools—Helicone, LangSmith, Langfuse, Future AGI, and Braintrust—for debugging hallucinated answers in production. The key requirement is the ability to quickly retrieve a compl…
Kennel 1.0.0, a native desktop app for managing multiple AI CLI agents side by side, has been released. The app allows users to run, organize, and persist sessions for tools like Claude Code and Kiro CLI, with sessions s…
Ethereum Foundation developer Austin Griffith launched a $1 AI-powered smart contract audit service on July 7, using the x402 protocol for microtransactions and ERC-8004 for onchain record-keeping. The service aims to ma…
IBM's guide on building AI agents explains the decision between using tools versus subagents for specific functionalities, emphasizing that tools execute deterministic code while subagents run separate LLM reasoning loop…
A developer released an interactive demo that visualizes how an AI agent thinks, showing its planning, actions, memory usage, and error recovery as it reasons about what to cook using ingredients in a fridge.
Pegasus One Health launched the SONG framework on July 7, 2026 to help healthcare organizations predict whether AI agents can scale from pilot to production. The framework evaluates Signal, Orchestration, Normalization, …
Anthropic has expanded its AI agent Claude Cowork to mobile and web platforms, previously limited to desktop. The agent operates in the background even when the laptop is closed, notifying users on their phone when a dec…
GitHub's Agentic Workflows, which use AI agents to automate tasks from plain-English instructions, are vulnerable to the GitLost indirect prompt injection attack. Noma Security demonstrated that an attacker can post a pu…
Synacktiv a dévoilé Pike, un agent LLM expérimental qui analyse les traces d'exécution de programmes Linux via une interface de chat. L'outil, construit avec Pydantic AI, permet de déboguer des crashes, identifier des ma…
Anthropic released the Model Context Protocol (MCP) in late 2024 to standardize how AI agents connect to data sources. A developer built a TypeScript filesystem server using the MCP SDK, demonstrating how to expose files…
NVIDIA released an AI agent built with NeMo and Nemotron models to automate industrial alarm analysis. The agent gathers context, runs specialist checks, and outputs structured evidence packages, aiming to reduce technic…
AI development is shifting from prompt engineering to loop engineering, where AI agents autonomously iterate and coordinate without human intervention. This transition transforms corporate governance, as loops create beh…
Anthropic launched Claude Cowork, a productivity tool within its Claude Desktop app, positioning it as a workplace assistant rather than a job replacer. The company softened its earlier warnings about AI-driven unemploym…
A developer traced the journey of a single LLM API call from a keystroke through submarine fibre cables, data centres, and busy GPUs, explaining why the same prompt can feel instant one day and sluggish the next. Another…
Cloudflare opened a July 1, 2026 waitlist for its Monetization Gateway, which lets customers charge per request for pages, datasets, APIs, and MCP tools using the x402 payment protocol. The service moves payment proof in…
Amazon Web Services introduced an AI-powered AWS Support Companion built on Amazon Bedrock AgentCore that consolidates incident investigation into a single conversational interface. The agent analyzes CloudWatch logs, se…
Motley AI achieved a 75.3% pass rate on the BIRD-INTERACT benchmark using the Claude SDK and its open-source semantic layer SLayer, far surpassing the official best of 36.33%. The agent harness provided the biggest impro…
INT21 has built SwarmOS, a cloud-native platform for multi-agent AI systems, arguing that orchestrating specialized agents is more effective than relying on larger context windows. The company demonstrated the approach b…