Articles, releases and code from Hacker News, Reddit, GitHub and the people building RPA, workflow automation and AI agents — plus what the community pushed to the top today.
AIUnited Airlines banned third-party AI agents from booking flights, restricting developers from automating reservations directly on its platforms and through travel aggregators.
AIDevelopers are using automated self-evaluations and custom workflow benchmarks to systematically test agent skills, mitigating erratic behavior across different underlying language models.
AIHackers used the open-source Hermes AI agent via Telegram to automate cyberespionage tasks, highlighting the critical need to secure autonomous execution environments against adversarial abuse.
AIRunning dozens of parallel AI coding agents accelerates SaaS development and backlog clearance, shifting the developer's role from writing code to reviewing output and resolving decision bottlenecks.
AIIBM launched agentic AI capabilities for mainframes, enabling developers to modernize, secure, and integrate legacy COBOL applications using the same automated workflows applied to modern languages.
AIEtherPK added automated version history and deleted item recovery to synced notes, allowing builders to safely let AI agents edit documentation without risking unrecoverable data loss.
AIDataiku released the open-source Kiji Inspector, using Sparse Autoencoders on decision tokens to help developers explain and monitor AI agents' tool selection choices in production.
AIMicrosoft released a local Azure CLI extension for Logic Apps Standard, allowing developers to build, run, test, and package workflows locally without Visual Studio Code.
AISwarms launched an API allowing users to integrate multi-agent AI teams directly into n8n, Zapier, and Make workflows using standard HTTP requests.
AIGoogle launched Gemini Decision 1, an API providing low-latency, single-pass classification and scoring across multimodal inputs to dramatically reduce routing costs in agent architectures.
AIEngineers are adopting fast, non-generative classification models like Jev directly within application code to execute deterministic scoring and automate decisions without costly, unpredictable LLM calls.
AIPlurai raised $13 million to test AI agents in simulated enterprise environments, helping builders detect edge-case failures and fine-tune cheaper models for reliable production workflows.
AICamunda 8 updated its backup and restore architecture, enabling decoupled, scheduled snapshots for relational databases to simplify disaster recovery for production workflows and AI agent actions.
AIAdobe announced Model Context Protocol support and background agents across Creative Cloud, enabling third-party AI agents to directly edit native project files and automate design workflows.
AIn8n is transitioning automation from centralized engineering teams to decentralized self-service models, requiring builders to adopt shared documentation, reusable templates, and enablement practices across departments.
AIGoogle open-sourced TPU Sync, accelerating KV cache and model weight transfers on TPUs to reduce inference latency for long-context models powering AI agents and automation workflows.
AILibreAGI launched a private BitTorrent tracker to archive open-weight AI models, helping automation developers access and preserve critical open-source models even if public hosts delete them.
AIAdobe launched Brand Intelligence, an AI platform that lets agents autonomously enforce brand guidelines, select layouts, and generate marketing assets directly within creative apps.