AI Products news and analysis on Web Pulse: 44158 curated articles tracking the latest AI Products developments, tools, and research, updated continuously from vetted sources.
Building software specifically for AI agents will mostly fail because tools that work well for agents also work well for humans, existing tools hold a training-data advantage, and the ideal ergonomics for agents remain u…
Cognition's SWE-2 coding model scored 83.75% on the eight-task KingBench 3 benchmark, beating DeepSeek V4.1 Flash at 81.25% and its own base model Kimi K3 at 77.5%, according to independent testing. SWE-2 is post-trained…
AI agents can safely customize existing Mac and Windows desktops when given scoped, limited control through tools with readable config files and documented commands, such as Aerospace on Mac, Apple Shortcuts, and PowerTo…
OpenAI released ChatGPT Images 2.5 alongside GPT-5 Astra, a computer-use agent that can operate a desktop across applications, and in a demonstrated run a single prompt carried two source photos through image generation,…
OpenAI released ChatGPT Images 2.5, an updated image generation and editing model available across desktop, mobile, and web on all tiers plus two API variants, adding sketch-to-image, improved identity consistency, and m…
OpenAI's ChatGPT Images 2.5, the successor to GPT Image 2, delivers its headline upgrade in frame-to-frame consistency, holding subject, style, and setting together across multi-step stop-motion edits far better than GPT…
AI agent harnesses — the tooling layer giving models file read/write and bash access — matter more to task completion than the underlying model, according to an analysis comparing Claude Code, Codex, and Hermes. The anal…
DeepSeek released V4.1 Flash, a 552-billion-parameter mixture-of-experts model with 8 billion active input and 16 billion active output parameters that scored 74.2 on Deep SWE and 30 on Terminal Bench 3.0, placing it nea…
DeepSeek released DeepSeek V4.1 Flash, a 552 billion parameter mixture-of-experts model under an MIT license that activates only 8 billion parameters during prefill and 16 billion during decode, cutting KV cache needs by…
TokenRhythm released NeoHorse-1-9B, a 9-billion-parameter model post-trained from Qwen3.5-9B that scores a 69.04 macro average across ten benchmarks, a 3.44-point gain over the 65.60 Qwen3.5-9B baseline. The Apache-2.0-l…
AI practitioners are advising developers to build model-agnostic agent skills — reusable instruction files, context documents, and plugins — so accumulated prompt and workflow expertise survives switching between harness…
Gartner, Inc. (NYSE:IT) posted an on-site Lead Software Engineer role in Gurgaon for NLP and generative AI, requiring 7-9 years of experience in algorithms, statistics, data mining, machine learning, deep learning and na…
Vercel said Sept. 10 that GitHub Copilot is now available in its AI SDK harness layer via the new @ai-sdk/harness-github-copilot package, which wraps the @ai-sdk/harness-acp adapter Vercel shipped Aug. 13 for the Agent C…
Digital Applied published a proposed maintenance responsibility matrix on September 12, 2026, to help buyers of AI-built applications assign a named owner for post-delivery upkeep. The framework separates incident respon…
Black Forest Labs' FLUX Video Edit endpoint, documented at /v1/flux-tools/video-edit-v1 and checked September 12, 2026, accepts MP4 URL or base64 source clips up to 15 seconds and 50 MiB, requires at least 160 pixels per…
Birgitta Böckeler's October 2025 analysis of spec-driven development tools Kiro, spec-kit and Tessl sorted approaches into three levels — spec-first, spec-anchored and spec-as-source — and found that all SDD approaches s…
Mbodi co-founder Xavier Chi said on TechCrunch's Build Mode podcast that robotics will not have a single ChatGPT-like breakthrough moment, arguing the value will accrue through incremental reliability rather than humanoi…
VideoRouter launched a unified API for video and image generation models that charges a flat 2% platform fee, compared with 5% on OpenRouter, claiming users can save up to 80% on video generations. The service provides o…
Microsoft engineer Stephen Toub opened nine pull requests from his phone at 35,000 feet on January 6, 2026, and seven merged, illustrating how coding agents let one person saturate a team's review capacity. A Firecrawl g…
Salesforce Inc. introduced a series of AI agents for sales and technical support teams, most of which are generally available today with the rest launching by year's end. The agents include Hunter, Piper and Carter for s…