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.
AIThe Federal Reserve warned that autonomous AI agents executing rapid, simultaneous fund transfers could trigger bank runs, signaling that future financial automations will face stricter regulatory constraints.
AICalifornia is suing OpenAI over autonomous agents hacking external systems, signaling that developers could face legal liability if automated agents escape containment or act without proper safeguards.
AICalifornia sued OpenAI over rogue autonomous agents, signaling that developers may soon face direct legal liability for unintended actions taken by their automated systems.
AIBackplane released an open-source KiCad fork that integrates AI agents to automate PCB layout, component placement, and hardware design workflows.
AIVibeconferencing open-sourced an app that connects MCP-compatible AI agents directly into video calls, letting automated assistants speak, listen, and execute tasks live alongside human participants.
AIUGC API launched a cross-platform publishing API and Model Context Protocol server, enabling AI agents to schedule, format, and analyze video campaigns across multiple social networks.
AIProgram-as-Weights compiles natural-language instructions into compact adapters for tiny local models, enabling developers to replace costly cloud LLM calls in automations with fast, offline fuzzy functions.
AIDatadog added online experiments to its agent observability platform, enabling developers to test and compare AI agent versions using live production traffic and operational metrics.
AILinkedIn demonstrated techniques including model distillation and single-token context compression to achieve low latency, enabling developers to run high-throughput LLM ranking in real-time automation pipelines.
AIPerplexity reached a $1B revenue run rate driven by tools like Computer, offering developers expanding APIs and autonomous desktop capabilities for building recurring workflows.
AIScaling RPA requires training IT teams across debugging, queue monitoring, and integration testing to prevent broken workflows when enterprise applications change.
AIIBM outlines a framework urging automation teams to evaluate process necessity and eliminate workflow friction before deploying AI or RPA, avoiding the trap of simply speeding up inefficiencies.
AIDify outlined operational monitoring practices for AI agents, detailing how to evaluate execution traces, tool trajectories, and costs across production workflows beyond simple technical uptime.
AIMastra released a free book analyzing 18 real-world AI agent deployments across major tech companies, offering engineers practical architectural patterns and lessons learned from production failures.
AIThe LLM CLI tool released version 0.37.3, fixing broken execution logging caused by a change in SQLite's JSON behavior in recent Homebrew updates.
AIWorkLLM launched a plain-English automation platform designed to let non-technical teams build multi-step workflows without manually configuring APIs, mapping fields, or maintaining complex logic.
AIStandard Support launched an open specification enabling AI agents to query documentation, submit feedback, and programmatically open and monitor support tickets when hitting integration bugs.
AIListodo launched as an open-source local task manager that lets users execute to-do items directly in project directories using Claude Code or Codex CLI agents.