Getting started with Luna
Luna is Ploy's AI identity agent: it answers questions about who has access to what, runs pre-built investigations, and builds automation from plain-English descriptions. This guide covers your first steps — where to open Luna, what each feature is for, and the path from a first conversation to a reviewed automation flow.
Before you start
Interacting with Luna is controlled by the Luna write permission in Ploy's permission system: "Interact with Luna, Ploy's AI assistant, to perform actions and get insights." If you can't interact with Luna, check that you have this permission (Understanding Ploy Permissions and Roles).
Luna works with the people, apps, and access data in your Ploy workspace, including your connected integrations.
Open Luna
Open the Luna page from the sidebar in your Ploy dashboard. The page is where you chat with Luna, run playbooks, and manage memories and artifacts.
Luna is also available as a chat bubble on every other page of the dashboard, and it uses the page you're on — a member profile, an app, or an access review — as context. You can attach files such as screenshots, CSV exports, and IdP reports to a conversation.
You can also talk to Luna from Slack once messaging is set up (Enable Luna Slack messaging).
What each Luna feature is for
Luna's features break down into six areas. Each links to the article that covers it in full.
Feature | What it's for | Learn more |
|---|---|---|
Chat | Ask plain-English questions about access, permissions, and risk. Luna answers in the chat with tables and cards, and can use the page you're on as context. | — |
Playbooks | Pre-built AI investigations with a goal, instructions, and a focused toolset. Default playbooks identify unmanaged apps, analyse access patterns, spot underused licences, and track down resource owners. | |
Memories | Facts about your organisation that Luna remembers across conversations and uses when answering questions, generating playbooks, and recommending access. | |
Artifacts | Luna's saved outputs — reports, summaries, analyses, investigations, and engagements — versioned and linked back to the chat that produced them. | |
Generated automation flows | Automation Luna builds from a plain-English description of a trigger, conditions, and actions. This is a beta feature — always verify a flow before activating it. | |
Access review suggestions | A recommendation for each account in a review — auto-approve, auto-reject, or needs review — which reviewers can accept or override. |
Your first Luna session
The safest route from first conversation to a reviewed automation: ask questions and run investigation playbooks first, then generate automation only when you're ready — and review everything Luna produces before it takes effect.
Open the Luna page from the sidebar.
Add the facts Luna should know on the Memories tab: your primary IdP, HRIS system, critical apps, and standard access patterns (Manage Luna AI persistent memory).
Ask a first question in chat — for example, "who owns this?" or "find anyone with similar access."
Run an investigation playbook such as Find Owner of Resource or Shadow IT Investigation to see how playbook sessions work (What are Luna playbooks?). The default playbooks investigate and report their findings rather than changing access directly.
Generate an automation flow once you know what you want to automate: use the Generate automation flow playbook and describe the trigger, conditions, and actions (Generate automation flows with Luna AI).
Revisit what Luna produced in the Artifacts tab, and open the source chat from any artifact (Artifacts).
Generated automation flows are beta and Luna may make mistakes. Review every flow's triggers, logic, actions, and scope before activating it, and test new flows with a narrow scope first.
When your first access review runs, Luna suggests a decision for each account — reviewers stay in control and can accept or override every suggestion (Article 8: Luna AI Suggestions).