We’re making Codex more useful for your work by expanding plugins beyond individual tools. These plugins turn Codex into a specialist for a specific role with a single install, no coding required. Codex can access 62 popular apps and 110 skills for work across sales, data
AGENTS
-
NVIDIA and Microsoft unveil unified stack for agentic AI at MSBuild
By
–
At #MSBuild, @Microsoft and NVIDIA unveiled an end-to-end stack for developers scaling agentic AI, from Windows devices to cloud to local. One unified stack. Every layer accelerated. Highlights include: RTX Spark: Windows PCs purpose-built for personal agents NVIDIA
-
LangSmith Sandboxes GA: Snapshots and cheap forks
By
–
New in the LangSmith Sandboxes GA Release: Snapshots and cheap forks Capture a running sandbox. Spin up 10 parallel branches for roughly the cost of one. When your agent goes down the wrong path, restore and try a different branch. https://
docs.langchain.com/langsmith/sand
box-snapshots
… -
Beyond moving agents: Durable platform, harness, and realistic dev environments are key.
By
–
A great cloud agent experience involves a lot more than moving a local agent to a server. We've learned that it requires a durable execution platform, a powerful harness, and the tools and infra to give agents realistic development environments.
-
Microsoft announces new Copilot super app with Autopilot agents and Scout
By
–
MICROSOFT : A new Copilot super app has been announced! It arrives with a concept of Autopilots, long-running, always-on agents, with Scout being the first Agent coming out of the box. More Autopilot Agents will be added later.
-
Microsoft scout reveals always-on personal agent for work
By
–
Microsoft scout revealed „your always-on personal agent for work.“ If "AI" was the Word of the Year in 2025, in 2026 it will be "agents" (always-on). Everything is agentic this year.
-
Agent Rubrics: Self-Correcting Output with Graders
By
–
New in Deep Agents: Agent Rubrics! Attach a rubric to your agent invocation, and a grader evaluates and self-corrects output until it satisfies all requirements. This is helpful for long/complex tasks where you need to keep the agent on track re an end goal!
-
Scout announced as first long-running Autopilot Agent
By
–
Scout is the first long-running Autopilot Agent!
-
Orchestrator agent dispatches sub-agents for entity research and verification
By
–
An orchestrator agent maps out which rows belong and where to find them, then dispatches sub-agents that each research one entity, verify it against real pages, and insert a row with sources attached. Schema inference runs on Claude Sonnet 4.6, the agents on Qwen3.7-max, all
-
TinyFish launches Bigset: open-source multi-agent system building verified datasets from plain language
By
–
TinyFish launched Bigset, an open-source multi-agent system that builds a verified, structured dataset from a single plain-language sentence, pulled live from the web and refreshed on whatever cadence you set.
— 🚨 AI News | TestingCatalog (@testingcatalog) 2 juin 2026
Describe the data, agents research it across real sources,… https://t.co/OLGmYKN5CE pic.twitter.com/XrRuCcVUteTinyFish launched Bigset, an open-source multi-agent system that builds a verified, structured dataset from a single plain-language sentence, pulled live from the web and refreshed on whatever cadence you set. Describe the data, agents research it across real sources,