All AI Pulse updates tagged "agents".
15 signalsOpenAI launched Patch the Planet, connecting security researchers, open source maintainers, and incident responders to coordinate disclosure and fixes when agents uncover critical vulnerabilities across open source.
Scale, Mercor and others now pay businesses for email, chat, tickets and code history to train agents on real office work; Secludy's free directory lists 159 such data sellers.
A Stanford and Together AI paper finds hybrid agents cut decoding compute by 62.5%–72.9% while scoring higher, suggesting most thinking works best in hidden vectors, not written text.
An attendee at LangChain Interrupt in London watched @hwchase17’s keynote on owning your intelligence, praising LangChain as a leader in open source and AI agents beyond GitHub and Microsoft.
The post highlights ICTRL as an open-source library letting computer or phone agents control iPhone apps, with the repo link in the comments.
The poster predicts 2027 will bring scaled autonomous revenue-generating agents or mini-companies, while noting today’s subscale versions remain rough.
Jason Lemkin frames a 2026 SaaS dilemma: agents do not need seats, pushing downgrades; raising prices, API fees, or AI charges each has limits.
The post claims OpenAI agents tried to recruit other models, over 1,200 agents communicated, and 70,000+ messages were exchanged; treat as unverified rumor.
tldraw product engineer Max shares how the offline local whiteboard launched in July lets users and their agents work together, discussing real-world usage and its limitations.
Author rwayne argues that as companies grow, they try to replicate early success and assume some magic is hidden in AGENTS.md, exploring why this is a misconception.
GithubProjects argues natural language is becoming the control layer for 3D work: not just AI-generated assets, but assets agents can understand, reason about, modify, and keep building with.
A Google paper says agents handle long tasks better when workflows live in editable procedure graphs that learn from execution, not buried in chat history.
Salesforce AI research finds that after tuning a weaker model’s prompts, tools, and workflow, copying a stronger model can hurt performance; targeted fixes work better.
The @aiworthusing x OpenClaw hackathon with @sodio asks builders to create a multiplayer agent that can do a real startup job, with Mac Minis and an SF trip as prizes.
Kimi K3 is now available on Amazon Bedrock for coding, document analysis, and extended agent workflows, with explicit prompt caching and AWS controls.