Skip to main content

πŸ€– AI Assistant

The AI Assistant answers member questions and moderates conversations in your server, grounding every answer in your Knowledge Base. You control when it responds, how it sounds, and what extra context it keeps in mind β€” all from the Fren One webapp.

Overview​

When configured, the agent can:

  • Answer questions automatically when members @mention it, reply to it, use a trigger keyword, or simply ask a question in chat
  • Ground its answers in your Knowledge Base and cite its sources
  • Speak with a personality you pick from presets or build yourself
  • Remember conversation context to improve later responses
  • Stay current through short Context Notes you maintain (launch dates, rule changes, announcements)

Configuration​

To configure the AI Assistant:

  1. Open your project in the Fren One webapp.
  2. Select the Discord server you want to configure.
  3. In the sidebar under Agent Configuration, select AI Features, then the AI Assistant tab.
  4. Toggle "Run the AI Moderator in this server" on.

Triggering​

Controls when the agent responds in chat.

OptionDescription
Trigger KeywordThe agent responds when this keyword is mentioned (e.g. !ask). Leave blank to always respond.
Respond to questionsAuto-detects questions in chat and answers them. Default: on.
Cooldown (s)Minimum seconds between agent responses, applied per member. Minimum 0; default 30.

Personality & Output​

Controls how the agent behaves and where it posts.

OptionDescription
Announcement Channel IDChannel for AI-related announcements.
Agent Personality (Soul)Markdown text describing the agent's personality and guidelines.
MemoryStores conversation context to improve responses. Default: on.

Personality presets and custom personalities​

Pick one of the preset personality cards under "Choose the assistant's personality preset", or build your own under Custom personalities with the Create button. The persona builder has:

OptionDescription
NameThe personality's name. Required; up to 100 characters.
ToneCasual, Professional, Technical, Friendly, Empathetic, or Authoritative.
StyleConcise, Detailed, Balanced (default), Step by Step, or Conversational.
LanguageAuto-detect (default) or a fixed language such as English, EspaΓ±ol, PortuguΓͺs, or Русский.
Custom instructionsOptional free-form guidance β€” what to emphasize, avoid, or how to greet. Up to 2,000 characters.

Context Notes​

Short notes that keep the assistant's context up to date with recent changes β€” faster to maintain than editing Knowledge Base sources for small facts. Select Add Note to create one; you can keep up to 20.

OptionDescription
TypeProfile, Rule, Preference, or Note.
KeyA short identifier, e.g. token_launch_date. Required; up to 64 characters.
ValueThe fact itself, e.g. "Token launches on April 15th". Required; up to 1,000 characters.

Memory Entries​

A read-only list of the conversation memory the AI has stored for this server. When nothing has been recorded yet, it shows "No memory entries".

How it works in Discord​

The agent responds in any channel it can read β€” no dedicated channel is required. It replies when a member:

  • @mentions the agent
  • Replies to one of the agent's messages
  • Uses the configured trigger keyword
  • Asks a question, when "Respond to questions" is on

Answers are grounded in your Knowledge Base, and replies cite where the information came from under "Sources:". The cooldown applies per member, so one enthusiastic member can't monopolize the assistant. Conversation memory is recorded even when auto-replies are off.

Members can also ask the agent directly with the /q question command, which posts a public answer generated from the knowledge base.

Saving your changes​

This page uses a batched save β€” edits stay local until you select Save in the bar that appears at the bottom of the page, and leaving with unsaved changes shows a warning. Context Notes are the exception: adding or deleting a note applies immediately.

Best practices​

Getting good answers
  • Fill your Knowledge Base before enabling the assistant β€” it can only answer from what it knows.
  • Use Context Notes for facts that change often (dates, prices, current events) instead of re-ingesting sources.
  • Keep the cooldown at 30 seconds or higher in busy servers so the agent doesn't dominate the conversation.
  • If the agent should only answer on demand, set a trigger keyword and turn "Respond to questions" off.
Troubleshooting

The agent isn't responding to questions?

  • Check that "Run the AI Moderator in this server" is toggled on and saved.
  • If a Trigger Keyword is set, the agent responds to that keyword, @mentions, and replies β€” check whether "Respond to questions" is also on.
  • The cooldown may be suppressing back-to-back responses from the same member.
  • AI features no-op silently when the AI service is unavailable.

Answers are missing information?

  • The assistant only answers from your Knowledge Base and Context Notes β€” add the missing content as a source or note.

On this page