Skip to content
Polo docs

What is Polo

Polo is an AI teammate for DevOps, platform engineering and SRE work. You give it a task the way you would ask a colleague. It plans the work, reads what it needs from the systems you connected, and proposes any change as one exact operation. A change runs only after a person approves it, or when an administrator has set a standing approval for that exact change. After a change runs, Polo checks the result with a separate read and reports what it found, what it changed and what the work cost.

The hosted product is at app.plungo.tech. You can also run Polo yourself.

Who it is for

Section titled “Who it is for”

Polo is for DevOps, platform and SRE engineers, and for teams that need that work done and want to review it rather than do every step by hand. Each company gets its own workspace. A workspace can have several agents, each with its own model, tools and limits.

What Polo does

Section titled “What Polo does”
  • Takes work in plain language. For example: “create the infrastructure for this service”, “investigate this outage” or “this pipeline is too slow”.
  • Plans and delegates. A supervisor, called Polo, reads each request and gives each part to the agent that holds the right tools. See How Polo works.
  • Reads your systems through connections. Cloud accounts, Kubernetes clusters, repositories, monitoring and alert sources are connected with the narrowest access that does the job. See Connections.
  • Proposes exact changes. A Terraform apply, a machine restart or a pull request is shown in full before it runs. See Approvals and guardrails.
  • Verifies and reports. Each change is checked with a separate read. The task shows the evidence, the artifacts and the cost.
  • Watches for trouble. Uptime monitors and Grafana alerts open incidents, and an agent can investigate them without being asked. See Incidents and uptime.
  • Runs recurring work. A weekly cost review or a daily backup check is a scheduled task. See Automation.

What Polo does not do

Section titled “What Polo does not do”
  • It does not change a system without approval. Everything outside an agent’s own workspace copy needs a person’s approval, or a standing approval an administrator set for that exact change.
  • It does not give itself more access. An agent uses only the tools switched on for it. A helper it creates cannot hold more than its parent.
  • It does not call an unverified attempt a success. When Polo cannot confirm an outcome, the task says so. A request whose answer was lost is shown as unknown and is not sent again.
  • It does not show unknown cost as $0. A model without a price is counted in tokens.
  • It does not cover every area equally. Each area of work has a maturity level, from planned to operated in a pilot. Open Settings → What Polo can do to see the level of each area and what your workspace has connected. See Capabilities.

How work reaches Polo

Section titled “How work reaches Polo”
Source What happens
Home and Tasks You write a task in the Ask Polo box or choose Assign task.
Slack and Google Chat A mention or a direct message to Polo opens a task.
Jira and ClickUp A ticket assigned to Polo opens a task.
Gmail An agent with a connected mailbox can read authorized mail.
Grafana alerts A signed webhook, or an alert bot’s message in a Slack alert channel, opens an incident.
Uptime monitors A failing check opens an incident and, if you choose, an investigation.
Automation A schedule creates a normal task at each run.
MCP clients and the terminal Claude Code, Codex, Cursor or the plungo command line work with a workspace through an access token.

Chat, ticket and alert sources are set up under Apps → Intake. Polo watches only the sources you connect. See Give work to Polo.

Next steps

Section titled “Next steps”