Pulumi Agents

AI agent · Workflow / back-office agent · IT & helpdesk by Pulumi

Managed AI agents for autonomous infrastructure operations and incident response.

Pulumi Agents are managed AI assistants that automate infrastructure tasks like incident investigation and drift remediation within Pulumi Cloud. Designed for Pro and Enterprise teams, they provide a way to scale infrastructure operations with human-in-the-loop controls and custom runbook integration.

Agent specs

The full spec card (context, pricing per token, modalities and more) on AntSpace. Sign up free to see the full spec card.

Category: AI Agents

Channels
Slack, Pulumi Cloud console
Deployment
Pulumi Cloud
Execution Environment
Pulumi Cloud
Supported Editions
Pro, Enterprise
Collaboration Platforms
Slack, Pulumi Cloud Tasks
Cloud Providers
AWS, Azure, Google Cloud
  • Managed AI agents running in Pulumi Cloud
  • Pre-built agents for incident investigation and drift response
  • Custom agent creation for specific team runbooks
  • Human-in-the-loop approvals for infrastructure changes
  • Integration with Slack for team collaboration

Overview

Pulumi Agents are managed AI agents that run within Pulumi Cloud to handle ongoing infrastructure responsibilities for engineering teams. They solve the problem of infrastructure bottlenecks by automating tasks such as incident investigation, drift response, and deployment troubleshooting. Teams can use pre-built agents for standard operations or build custom agents tailored to their specific runbooks and operating procedures. These agents operate with saved instructions and connected tools, such as logs and code repositories, while remaining under human control through mandatory approvals and role-based access. By integrating with Slack, Pulumi Agents allow developers to collaborate on infrastructure fixes and investigations in real-time.

Operational Tasks

  • Incident Investigation: Agents correlate application errors with cloud configurations and recent deployments to identify root causes.
  • Drift Response: Managed agents monitor for infrastructure drift and can initiate automated responses to maintain the desired state.
  • Deployment Troubleshooting: Agents investigate deployment failures and prepare fixes for infrastructure issues to be reviewed by humans.
  • Dependency Maintenance: Automated agents handle the ongoing maintenance of infrastructure dependencies to keep systems up to date.

Agent Management & Control

  • Custom Agent Creation: Users can build agents tailored to their team's specific infrastructure and operating procedures, such as incident runbooks.
  • Tool Connectivity: Agents connect to cloud resources, code repositories, logs, and issue trackers to perform their work.
  • Automated Work Assignment: Work can be assigned directly to agents or triggered automatically based on events and schedules.
  • Managed Cloud Execution: Agents run in Pulumi Cloud with saved instructions, connected tools, and specific permissions for their work.

Collaboration & Governance

  • Slack Integration: Users can collaborate with agents in tasks or Slack threads to add context and refine the agent's work.
  • Access Control: Organizations control which stacks, cloud accounts, and repositories an agent can access.
  • Approval Workflows: Human-in-the-loop approvals ensure that agents only execute changes after a person has reviewed them.

What it does

  • Investigating incidents\nResponding to infrastructure drift\nTroubleshooting deployments\nMaintaining dependencies\nPreparing infrastructure fixes for review

Use cases

  • Infrastructure teams use agents to correlate application errors with cloud configurations to speed up incident root cause analysis.
  • Platform engineers deploy drift-response agents to automatically detect and propose fixes for unauthorized infrastructure changes.
  • DevOps teams build custom agents that follow specific company runbooks to maintain dependencies across multi-cloud environments.
  • Service owners collaborate with agents in Slack threads to troubleshoot failed deployments and approve AI-generated infrastructure fixes.

Specifications

Channels
Slack, Pulumi Cloud console
Integrations
Datadog, PagerDuty, Linear, GitHub, GitLab, Azure DevOps, AWS, Google Cloud, Microsoft Azure
Human control
Human-in-the-loop approvals for changes, collaboration in Slack threads, and role-based access controls for repositories and cloud accounts
Deployment
Pulumi Cloud
Setup
Zero-setup for pre-built agents like Neo; custom agents require connecting tools and providing instructions
Execution Environment
Pulumi Cloud
Supported Editions
Pro, Enterprise
Collaboration Platforms
Slack, Pulumi Cloud Tasks
Cloud Providers
AWS, Azure, Google Cloud
Integration Protocol
MCP (Model Context Protocol)

How it works

  1. Choose or Build an Agent: Select a pre-built agent for common tasks or create a custom agent by providing specific instructions and runbooks.
  2. Connect Tools and Permissions: Grant the agent access to the necessary cloud accounts, code repositories, and monitoring tools like Datadog or PagerDuty.
  3. Automate Workflows: Assign tasks directly to the agent or set up triggers based on schedules or system events like deployment failures.
  4. Review and Approve: Review the agent's findings or proposed fixes in Pulumi Cloud or Slack, and provide approval before changes are applied.

FAQ

Which Pulumi Cloud plans include Pulumi Agents?

Pulumi Agents are available in the Pro and Enterprise editions of Pulumi Cloud.

What tools can I connect to Pulumi Agents?

You can connect agents to your cloud resources, code repositories, logs, and issue trackers so they can perform tasks like incident investigation.

Can I create my own custom agents?

Yes, you can build custom agents tailored to your team's specific infrastructure, operating procedures, and incident runbooks.

How do I control what an agent can do?

You control which stacks and cloud accounts an agent can access, and you can require human approval before an agent executes any changes.

Added by AntSpace from public sources (pulumi.com).

Related

More AI agents by Pulumi