Anthropic ships Claude Opus 4.1, claims state-of-the-art on real-world coding
Opus 4.1 posts a 74.5% score on SWE-bench Verified and introduces a new “computer use” beta that lets agents click, type and scroll inside live applications.
Anthropic
Agents
AI Tools
Perplexity launches Comet, an AI-native browser built around an agent
Comet replaces tabs with tasks and ships with an always-on agent that can buy, book and research on your behalf. We took it for a spin.
AI Tools
Hugging Face’s `smolagents` proves you can build an AI agent in 100 lines
A minimalist agent framework from Hugging Face is quietly becoming the default for teams who got tired of LangChain abstractions.
Automation
LangChain’s LangGraph Platform goes GA with managed agent orchestration
After 6 months of preview, LangGraph Platform is now generally available with autoscaling, durable runs, and a visual debugger that actually works.
Automation
Stripe ships an Agent Toolkit so AI agents can actually move money
A new SDK from Stripe lets AI agents create invoices, charge cards and issue refunds — with first-class safety controls.
AI Tools
GitHub Copilot Workspace can now finish entire issues end-to-end
Assign a GitHub issue to Copilot, walk away, come back to a working pull request. We tested it on 25 real bugs.
AI Tools
Replit Agent v3 ships full-stack apps from a single prompt
Type a sentence, get a deployed app. Replit’s newest agent revision is the closest thing yet to truly no-code AI development.