AI Agents news and analysis on Web Pulse: 13199 curated articles tracking the latest AI Agents developments, tools, and research, updated continuously from vetted sources.
OpenAI's GPT-5.6 preview is appearing in Codex and API surfaces for approved organizations, but remains unavailable for broad self-service use. Builders must address the launch gap—including stable URLs, access control, …
Nocta released activity-frames, an open-source tool that records a user's screen locally and compiles it into structured activity frames for AI agents, enabling episodic memory without cloud processing or LLM inference. …
Compound Engineering (CE) released /ce-dogfood, a QA agent skill that autonomously tests code changes in a real browser by diff-scoped, flow-based verification, closing the verification loop without human intervention. T…
A live AI chatbot built on Groq's API scored 55/100 (Grade E) in a stress test using BotCritic, revealing critical failures including invented identity, fabricated statistics, and inability to handle context. The bot fal…
Anthropic released Claude Opus 4.7 on April 16, 2026, a targeted upgrade focused on making agentic coding loops and multi-step tool use production-ready. The model achieves 87.6% on SWE-bench Verified and 64.3% on SWE-be…
Z.ai's open-weight GLM 5.2 completed an agentic coding task in 17 minutes at $2.76, while Anthropic's Claude Fable 5 finished in 9 minutes at over $10, with comparable output quality. The open model cost roughly a quarte…
Ambarella is using AI agents to abstract its software development kit, enabling independent software vendors to port applications to its hardware in days instead of months. The company has onboarded ISVs in healthcare, r…
A new study from the barkup-bench series reveals that LLM agents suffer from session drift when serializing tree state only once, with per-step success dropping from 98.8% to 83.8% for Claude Sonnet 4.5 over 12 turns. Th…
Yggdrasil Agent, a general-purpose parallel reasoning system, launches with a multi-agent architecture that splits work across ten cognitive realms to improve accuracy and reduce blind spots. The system is vendor-agnosti…
A developer consolidated four repositories by eliminating duplicate sources of truth. They migrated a SaaS from per-feature subscriptions to plans-only entitlements without billing outage, unified authorization across we…
NVIDIA positioned its Vera CPU for agentic AI workloads, claiming it outperforms x86 in coding tasks by up to 1.9x. The company highlighted Vera's Olympus cores and memory bandwidth as key for CPU-bound agent loops. NVID…
AlpineGate AI Technologies launched AgentFactory, a governed Digital Intelligence platform that converts plain-language business objectives into reviewable WorkOrders, agent teams, approvals, evidence, and deliverables. …
A developer shared a 15-minute review checklist for agent-generated pull requests to prevent technical debt and scope drift, including checks for edge cases like empty input and hidden refactors. The process catches issu…
Rumik introduces Ira, a voice AI that combines three core components: Silk for expressive voice, Mesh for memory management, and Peek for contextual conversation understanding.
Nitro Software launched Nitro Automate, an intelligent document automation platform that combines document processing, workflow automation, and AI-powered integrations to eliminate manual document tasks in enterprise wor…
Backlog, a local-first task and context manager for AI coding agents, has been released as an open-source tool that stores tasks, plans, and memory in a single SQLite file shared between humans and AI agents. The tool at…
Evokoa launched Polygres, a managed PostgreSQL platform that integrates native graph traversal, HNSW vector search, and hybrid retrieval APIs to enable agents to query structured records, relationships, and semantic mean…
Anthropic published a post-mortem on Claude Code degradation, tracing quality issues to compounding changes including a reasoning effort downgrade and a caching bug. GitHub moved Copilot to usage-based billing as agentic…
The Linux Foundation introduced the Agent Name Service (ANS), an open, DNS-inspired standard for giving AI agents verifiable internet-native identities. ANS anchors agent identity to DNS, enabling systems to verify who a…
Meta launched Muse Image, its first dedicated AI image-generation model, across Instagram, WhatsApp, and the Meta AI app, built by its Superintelligence Labs division. The agentic tool works with the Muse Spark language …