OpenAI Codex Developer Warns Against AI Agent Swarms
OpenAI Codex developer Matthias Bastian warns that using more than two parallel sub-agents wastes tokens without improving quality.
676 articles
The tools, frameworks, and platforms developers use to build with AI — coding assistants, agent frameworks, APIs, and developer tooling, with the updates that change how you ship.
OpenAI Codex developer Matthias Bastian warns that using more than two parallel sub-agents wastes tokens without improving quality.
Anthropic has introduced parallel agent workflows in Claude Code, allowing users to split coding tasks across multiple threads for improved efficiency.
Amazon SageMaker AI and Rekognition generated synthetic data improved person detection mAP50 by 160 percent in industrial safety AI, according to a recent blog post.
Amazon announced a new authorization framework for MCP tools on Amazon Quick, designed to enforce granular access controls with four sequential gates.
Wood Mackenzie released APEX, a shared agentic platform, to address challenges in scaling AI agents, with 88% of proofs-of-concept failing to reach production.
AWS announced a serverless Git metrics dashboard using Amazon QuickSight, offering near-real-time analytics for development activity.
HuggingFace released Funes, a local memory system for coding agents, on September 17, 2026, allowing agents to recall past sessions without relying on cloud-based models.
Hugging Face released hfdask, a tool that enables synthetic data generation using Dask on Hugging Face Jobs, on September 15, 2026.
OpenAI introduced new analytics tools on September 16, 2026, enabling admins to track AI usage, costs, and business outcomes across ChatGPT Work and Codex.
AWS announced Amazon Bedrock Data Automation, a tool for automating PII redaction in documents, capable of processing thousands of scanned files daily.
AWS announced a new method for fault-tolerant distributed training using NVRx on Amazon EKS, reducing idle time by up to 40% in large-scale jobs.
AWS released 38 open-source agent skills for healthcare and life sciences (HCLS) AI reasoning, improving decision accuracy by up to 86% in head-to-head comparisons.