Agents
Long-running AI agents that plan, draft, and execute work for you across Flamel.
For everyone. Agents run inside the same scope you're already in: as a workspace user, they act on your workspace's data; as a Hub admin, they can act network-wide.
Agents are conversational AI sessions that go beyond Luna AI. Where Luna answers a question or drafts a single piece of copy, an Agent can plan a multi-step task, call Flamel tools, and check back with you between steps. Think of Luna as a quick reply and Agents as a focused work session.
When to use Agents instead of Luna AI
| You need… | Use |
|---|---|
| A caption, hashtag list, or quick rewrite | Luna AI |
| A whole month of posts drafted, scheduled, and tagged | Agents |
| A bulk audit of a workspace's connected accounts | Agents |
| Pulling Analytics data and turning it into a recap | Agents |
| Researching competitors and proposing post ideas | Agents |
How a session works
Start a session
Click Agents in the sidebar → New Session and describe what you want done in plain English.
Let it plan
The Agent breaks the request into steps and either runs them or asks you to confirm risky ones (sending posts, opting into a paid campaign, deleting media).
Step in when needed
You can pause, redirect, or hand the Agent a file or link mid-session. Conversation history is preserved per session.
Review and ship
The Agent shows what it did, what it drafted, and what's pending your approval. You decide what to publish.
Sessions and scope
Each session is scoped to the context you started it in. Switching workspaces or hubs starts a new session — the Agent never sees data from a scope you can't see yourself. Sessions persist so you can come back to an in-progress task tomorrow.
What Agents can do today
- Draft and schedule organic posts across connected platforms
- Reply to comments and DMs using your Brand Kit voice
- Pull spend, ROAS, and engagement numbers and write a recap
- Generate and refine Templates
- Triage Google Reviews and propose replies
Capabilities expand frequently. If you wish an Agent could do something it can't yet, tell it — feedback routes directly to the roadmap.