$ today_in_10_seconds
Virtana just put its ops agents inside Slack, Teams and Claude. You ask why prod is slow, you get a cross-stack diagnosis, and you approve the fix without leaving the thread. The signal is narrow but clear: agents are moving out of dashboards and into the channels where humans already argue about incidents.
$ lead_story
Virtana Brings Agentic Observability to Microsoft Teams, Slack and Claude
Virtana announced integrations with Microsoft Teams, Salesforce Slack and Anthropic Claude that let engineers and IT leaders work with its AI agents in natural language. Its own example goes like this. An engineer asks in Slack why the order processing app is slow, and the agents trace the problem across services, databases, compute, network and storage, using Virtana's System Dependency Graph and more than 25,000 operational metrics. They identify a database bottleneck caused by a recent deployment, and Virtana then coordinates specialized agents to run the fix and confirm performance has recovered. The human's job shrinks to reviewing and approving. One caveat: it's available now only to select customers, with general availability planned for Q1 2027.
// why_it_matters: The approve button is moving from the ops console to the chat thread, and whoever owns the context graph behind it owns the incident.
$ also_worth_attention
Virtana's agentic war room: MCP + A2A for incident response
Today's chat integrations sit on top of this earlier launch. Virtana's Agentic Observability Enterprise Management uses Model Context Protocol and the open Agent2Agent protocol to recruit agents from third-party platforms into a shared incident war room, citing Salesforce research that the average enterprise runs 12 AI agents and half operate in silos. The chat UI is the front door. The real bet is this orchestration layer behind it.
link →Claude Tag: Anthropic's own agent-in-Slack
Anthropic's Claude Tag works under its own identity, builds context from the channels it's in, and can follow up on its own. It's in beta on Team and Enterprise plans. It's Slack-only for now, and Teams support is a waitlist with no date. Virtana shipping Teams on day one highlights that Teams is where a lot of enterprise ops lives, and it's still an open lane.
link →Virtana's MCP Server: the plumbing under today's news
Back in February, Virtana exposed its dependency graph through MCP so agents built on ChatGPT, Claude and Gemini could query structured operational context directly. That's why a chat integration could follow within months. If your product's context isn't reachable over MCP, agents can't see it.
link →
$ bigger_picture
The trend behind the headlines
Chat is quietly becoming the control plane for agents. You can see it in Anthropic's Claude Tag in Slack, in open-source projects that drop any agent into a channel, and now in observability vendors routing remediation through a thread. The interface itself is turning into a commodity. What sets products apart is the context graph the agent reasons over and the governance around approve-to-act: who approved what, with which permissions, and with what audit trail. MCP and A2A mean your agent will increasingly be recruited into someone else's room. So builders face a strategic choice: be the specialist tool that gets called, or be the orchestrator that does the calling.
SandboxAQ Switch
This open-source project turns Slack, Microsoft Teams and Discord channels into shared rooms where people and agents work with the same context and history. It supports agents built with Claude Code, Google ADK, LangChain, OpenAI and more. It's a good way to prototype chat-native agent workflows before you commit to a vendor's version.
$ subscribe --free
Get the next edition in your inbox
// DAIly Agent covers agentic AI three times a week
// free, no spam, exit anytime