AI Infrastructure news and analysis on Web Pulse: 35262 curated articles tracking the latest AI Infrastructure developments, tools, and research, updated continuously from vetted sources.
Miodrag Vilotijević, co-founder of JigJoy, argues that AI agents remain trapped inside the projects that created them, preventing reuse and specialization, and proposes three architectural pillars—concurrency, awareness,…
Io.github.0rkz/byte-protocol version 0.12.3, a per-byte data protocol for AI agents using x402 USDC on Base with EIP-712 attestation, was published to the official MCP registry on 2026-09-03. The server, distributed as b…
Io.github.SAIHM-Admin/saihm-mcp, an enterprise-ready portable memory server for AI agents, was published as v0.3.13 to the official MCP registry on 2026-09-03, distributed as @saihm/mcp-server on npm and filed under Memo…
A developer's analysis of production AI agent failures reveals that silent failures—where systems complete execution without errors but produce incorrect or harmful outputs—are the dominant failure mode, often slipping t…
A developer who built an autonomous Claude Code sales setup generating ¥1.2M a month reported that their macOS LaunchAgents were rewritten to send outreach DMs at 2:31 AM, risking account flags. They created a self-heali…
Supabase will retire its logs.all Management API endpoint on September 23, 2026, replacing it with an endpoint that only accepts ClickHouse SQL, according to a changelog post published July 23. The change breaks AI agent…
Fly.io has launched Sprites, disposable cloud computers designed specifically for running AI agents, with a new MCP server and plugins for Claude Code, Codex, Cursor, Antigravity, opencode, and Grok that enable agents to…
A 1Password survey of technical workers found that 62% reported gaps in how their company manages AI agents, highlighting the challenge of controlling AI costs as usage-based billing drives up expenses. The article outli…
Anthropic launched Claude Fable 5.1 and Mythos 5.1 on September 1, featuring a 1M-token context window by default, always-on adaptive thinking, and prompt cache reads reduced to $0.25 per million tokens. Salesforce's Win…
DeepSeek-AI released the DeepSeek-V4 series on April 24, 2026, including the MIT-licensed DeepSeek-V4-Flash model with 284B total parameters (13B activated) and a one-million-token context window, and a blog post details…
Elastic and Tencent Cloud announced an expanded strategic partnership in Shenzhen in August 2026, launching Tencent Cloud Elasticsearch Service (ES) Enterprise Edition to power AI-era search. The collaboration integrates…
Inco AI launched its inference platform in public beta, claiming the fastest output speeds on Artificial Analysis leaderboards for four open models: Kimi K3, MiniMax M3, GLM 5.3, and GLM 5.3 Flash. The company says its s…
OpenAI, Nvidia, and Anthropic are pursuing distinct AI compute strategies: OpenAI revealed its custom inference chip Habanero, claiming it beats Nvidia's GB200 and GB300 systems on latency and throughput per kilowatt, wh…
OpenAI, Nvidia, and Anthropic are pursuing divergent strategies in AI infrastructure, with OpenAI vertically integrating by developing its own inference chip (code-named Habanero) and cutting off future model access to C…
A guide to shipping a Model Context Protocol (MCP) server for a product reports that building a compliant server is a short job under the current SDKs, but distribution is a long tail of directory submissions, with one s…
Ben Houston's Three-LLM library runs LLMs entirely in the browser via Three.js and WebGPU, supporting GPT-2, Llama-style, Gemma 3, Phi, and Qwen3.5 architectures, with checkpoints from 3M to 1.3B parameters. The open-sou…
OpenAI's GPT-6 Astra release on September 3, 2026, introduced asynchronous tool calling in its Responses API, allowing models to continue generating while client-side tools run, a capability not offered by Anthropic or G…
Data platforms including Snowflake, Databricks, ClickHouse, BigQuery, MotherDuck, and Redshift are embedding model inference and agents directly into query execution, according to an analysis of AI architecture in the da…
SK hynix, in a series on the AI ecosystem, argues that the primary bottleneck for AI performance is shifting from compute power to data movement and memory architecture, particularly as AI workloads move toward inference…
Field AI, a robotics company based in Irvine, California, is hiring a Senior/Staff Software Engineer for its Robotics Platform team to build the software platform that enables its foundation models to run reliably across…