> ## Documentation Index
> Fetch the complete documentation index at: https://sippet.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Set up Discord support monitoring and replies

> Connect Discord, bind support channels, assign an AI agent, and review examples before scaling replies.

## Goal

Connect a Discord workspace and let an AI agent monitor and respond to product questions in selected channels.

## Before you start

* You should have a published AI agent ready for support use.
* Decide which channels are safe for AI-assisted replies.
* Decide whether the agent should be conservative or proactive.

## Step 1: connect the workspace

In the Discord integration area:

* install or connect the Discord app
* create the workspace binding
* confirm the workspace shows as active before continuing

## Step 2: bind the channels

For each support channel:

* add a channel binding
* choose whether listening is enabled
* choose whether replies are enabled
* decide whether channel history should be backfilled

## Step 3: assign the agent

Attach a support-ready AI agent to the workspace binding so replies use the right prompt, tools, and escalation behavior.

## Step 4: choose the reply mode

Use a narrower mode first:

* `mention or question` for safer rollouts
* broader reply behavior only after you trust the agent in that channel

## Step 5: seed and review examples

Use examples to:

* keep tone aligned with your support style
* reinforce good answers to recurring questions
* review whether the agent should answer, escalate, or stay silent

## Example workflow

* a user asks a product question in a support channel
* the AI agent reads the question in Discord
* the agent answers using its configured knowledge and support instructions
* if the case is ambiguous or sensitive, the flow escalates to a human instead of over-answering

## What good looks like

* replies are accurate and consistent
* the agent only speaks in the channels you explicitly enabled
* examples improve quality over time instead of being ignored

## Troubleshooting

* If replies are too broad, tighten the reply mode.
* If the agent misses context, add examples or connect more product knowledge.
* If a channel is too noisy, start with observe-only behavior before enabling replies.

## Related pages

* [Discord integration overview](/console/discord-integration)
* [Connect GitHub knowledge to an AI agent](/console/github-integration-guide)
* [Create a no-code support agent](/console/ai-agents-guide)
