Lumo's Marketing Agent autonomously plans, prepares, and analyzes toward your goals, so you can run everything simply by setting goals and approving.
What is the difference between the Marketing Agent and the Customer Agent?
They support different people. The Marketing Agent supports the Members of your Workspace, while the Customer Agent automatically responds to your customers on LINE.
| Agent | Audience | Main role |
|---|---|---|
| Marketing Agent | Members who use the management screen | Autonomously plans, creates, and analyzes broadcasts |
| Customer Agent | Customers on LINE | Automatically responds to inquiries from friends based on the Knowledge Base |
Will the agent ever send messages on its own?
No. Sending a broadcast always requires your approval.
The agent autonomously handles planning, creating, and scheduling a broadcast, but the final send happens only after you select "Approve" in the approval dialog. If you do not want to send it, select "Cancel". A broadcast left unapproved will never be sent automatically.
What is the Trust Score?
It is a mechanism that manages how far the agent may proceed automatically, separately for each type of operation. In the management screen it appears as "Trust in the AI agent."
The more proposals you approve, the higher the Trust Score for that type of operation rises, so tasks like drafting and creating Segments begin to proceed automatically. Depending on the score, an operation moves to the next stage.
| Level | Agent behavior |
|---|---|
| Approval required | Always asks for approval before acting |
| Notification only | Notifies you of the action but does not ask for approval |
| Automatic | Acts automatically without confirmation |
However, sending a broadcast and activating a Workflow always require approval, regardless of the Trust Score.
What should I do if I feel I am getting too many notifications from the agent?
You can dismiss notifications, and they are tuned so that the same type of notification does not arrive repeatedly within a short interval.
Unwanted notifications are hidden once you dismiss them. Because there is a mechanism that makes it harder for the same type of notification to arrive consecutively, similar notifications are kept from coming in quick succession.
How does the agent learn?
It learns automatically from broadcast results and from your edits. What it learns is reflected in proposals that fit your brand.
| What it learns | Source of the learning |
|---|---|
| Broadcast tone | Past broadcast text and your edits |
| Broadcast timing | Performance data such as open rates |
| Segment effectiveness | Which Segments responded to which themes |
| Your preferences | Patterns in your edits and feedback |
You can review and delete the learned memory under AI agent in Personal Settings.
Are there additional charges for using the Marketing Agent?
As long as you stay within the scope included in your current plan, there are no additional charges.
The available scope differs by plan. Check the Plans & Billing page for the detailed limits.
What should I do if the agent makes a proposal that differs from my intent?
Just tell it the direction directly in Chat. The agent adjusts its proposal based on the context of the conversation.
You can give corrections in natural language, such as "Make it more casual" or "Narrow the audience more." Sending feedback from "Thumbs down" shown beneath a response helps improve the accuracy of future proposals.
How do I delete my chat history?
There is a way to delete an individual chat, and a way to clear all history and memory at once.
Delete an individual chat
From the conversation history in the sidebar, select the chat you want to delete and delete it.
Clear all history and memory
Run "Clear conversation history and memory" under AI agent in Personal Settings. All conversation history and memory between you and the agent are deleted, and this action cannot be undone. It does not affect other Members' history.
Can I change the way the AI talks?
Yes. Under AI agent in Personal Settings, choose a personality from the presets.
The personality you choose applies only to your own Chat and does not affect other Members.
| Personality | Speaking style |
|---|---|
| Empathetic | Responds with reassuring phrasing that acknowledges your situation and feelings |
| Friendly | Responds in an approachable, easy-to-talk-to tone |
| Polite | Responds in a calm, business-oriented tone, organized and easy to understand |
| Simple | Responds by summarizing the conclusion and key points briefly and concisely |
If you want to use the Workspace's standard setting as is, select "Follow default settings".
What is the difference between the chat modes?
There are three modes, and the focus of the agent's responses changes. If you are unsure, we recommend Auto.
| Mode | How it works |
|---|---|
| Auto | Automatically switches between proposing and creating depending on the situation |
| Create | Prioritizes creating starting points and drafts, such as broadcasts and Workflows |
| Consult | Prioritizes brainstorming, analysis, and clarifying direction |
You can switch modes at any time from the chat screen.
Can I use it without setting a goal?
Yes. Even without a goal, you can use the full range of the agent's features, such as creating broadcasts and consulting on analysis.
When you set a Goal, all of the agent's proposals become tied to achieving that goal, and features like building a plan and checking progress come into their own. You can set a goal even with a vague expression like "I want more friends," and the agent breaks it down into concrete metrics.
Where can I change the personality setting?
Open "Personal Settings" from the profile icon in the navigation menu, and choose a personality under AI agent.
You can choose the personality from presets such as "Empathetic," "Friendly," "Polite," and "Simple." The personality you choose applies only to your own Chat and does not affect other Members.
Are tags applied in Chat different from Friend Tags?
No — they are the same. A tag is a single kind of label attached to friends. Whether you apply it from the friend info panel in Chat (the inbox), from the friend list, or from the tag management screen, it is treated as the same tag.
| Type | Purpose |
|---|---|
| Tag | An attribute label attached to each individual friend. Used as a condition to narrow down a broadcast |
| Recipient Segment | A filtering condition that combines criteria such as tags. Used as the recipients of a broadcast |
When you discuss broadcast recipients, tags and Segments are what matter. Tags you apply while handling chats can be used directly to narrow down a broadcast.