How do you add an AI bot to Discord?

There are three realistic ways to put an AI bot on a Discord server. The first is to invite a hosted bot that already includes AI: MEE6's AI plugin, for example, adds a /write command powered by OpenAI technology, but it needs a paid MEE6 AI subscription and you cannot pick the model. The second is to run your own bot: create an application in the Discord Developer Portal, turn on the Message Content intent, and connect its bot token to software that calls the model you choose, such as Claude, GPT or Gemini, with your own API key. The third is the same self-run bot pointed at a local model through Ollama, so prompts never leave your computer. OpenClaw Easy is a free desktop app for macOS 12+ and Windows 10/11 that sets up the second and third options without code: paste a bot token and Server ID, pick a model, and the bot answers @mentions and direct messages while the app runs.

Discord does not ship its own AI assistant. It tested an experimental chatbot called Clyde, built with OpenAI technology, and shut it down on December 1, 2023. So every AI bot you see on Discord today is either a third-party bot someone invited or a bot a server owner runs themselves. This guide compares those options honestly, then walks through the no-code way to add an AI bot to Discord with OpenClaw Easy (version 2026.10.1).

OpenClaw Easy Quick Actions dashboard with the assistant online

3 Ways to Get an AI Bot on Discord, Compared

The right choice depends on four things: who holds your server's messages, which model answers, what it costs, and how much you can customize. Here is how the three approaches stack up.

Your own bot + API key Your own bot + local model Hosted AI bot (invite link)
Which model answers Any model you have a key for: Claude, GPT, Gemini Open models you download with Ollama (Llama, Qwen, Mistral and others) Whatever the bot vendor chose
Where messages go Your computer and the AI provider you picked Stay on your computer The vendor's cloud, then its AI provider
Cost Provider token pricing, pay per use Free to run; needs capable hardware Usually a paid tier for AI features
Customization Full system prompt, model choice, fallbacks Full system prompt, model choice Limited to the vendor's dashboard options
Uptime While your machine runs the bot While your machine runs the bot Vendor-hosted, always on
Setup About 10 minutes with OpenClaw Easy Same, plus installing Ollama and a model Click invite, authorize

Option 1: Invite a hosted AI bot

Hosted bots are the fastest start: you click an invite link, pick your server, and the bot is there. Community-management bots have added AI on top of moderation and leveling. MEE6 is a common example: its AI plugin offers /write for text answers (powered by OpenAI technology, per MEE6's own documentation), /imagine for images, and AI characters with custom backstories, and it requires a MEE6 AI subscription. The trade-offs are the same across hosted bots: you do not choose the underlying model, your server's messages pass through the vendor's infrastructure, and the AI behaves the way the vendor configured it. For a side-by-side of hosted picks, see our best Discord AI bots of 2026 roundup.

Option 2: Run your own AI bot with your own API key

Here you own the bot. You create it in the Discord Developer Portal, so it has your name and avatar, and you connect its token to software that sends mentions to a model of your choice. Traditionally that meant writing a discord.js or discord.py bot and renting a server to host it. The benefits are control and choice: pick Claude for long, careful answers or GPT or Gemini for general knowledge, write your own system prompt, and pay the AI provider only for the tokens you use. The costs are that you manage an API key, and a self-run bot is only online while the machine running it is on.

Option 3: Run your own AI bot on a local model

The same self-run bot can call a model on your own hardware. Ollama, an MIT-licensed tool for macOS, Windows and Linux, downloads and runs open models locally, so no prompt or reply is sent to a cloud AI provider. This is the most private and cheapest option once set up. The trade-offs: local models are generally less capable than the best hosted models, and answer speed depends on your machine. 16 GB of RAM or more is a sensible starting point.

Where OpenClaw Easy fits: it covers options 2 and 3 without code or a rented server. You bring a Discord bot token and either an API key, OpenClaw Easy pre-paid credits, or a local Ollama model. The app runs the bot on your Mac or PC.

What You Will Need

  • A computer running macOS 12 or later (Intel or Apple Silicon) or Windows 10/11.
  • A Discord account that can add apps to the server (server owners and admins can).
  • An AI model: an API key from Anthropic (Claude), OpenAI (GPT) or Google (Gemini), OpenClaw Easy pre-paid credits, or a local model through Ollama.
  • About 10 minutes.

Step-by-Step: Add an AI Bot to Discord with OpenClaw Easy

1 Download OpenClaw Easy

Go to the download page and get the installer for your OS:

  • macOS: .dmg file (Apple Silicon or Intel), Developer-ID signed and notarized by Apple.
  • Windows: .exe installer. It is not code-signed yet, so Microsoft SmartScreen warns on first run; click More info, then Run anyway.

Install and open the app. There are no dependencies to install and no terminal commands.

2 Create a Discord Bot and Copy Its Token

  1. Open the Discord Developer Portal (discord.com/developers/applications) and click Create App. Give it a name such as "AI Assistant".
  2. Open the Bot page in the sidebar and click Reset Token. Copy the token right away; Discord will not show it again unless you reset it.
  3. On the same page, under Privileged Gateway Intents, turn on Message Content Intent so the bot can read the messages it answers.

Tip: Anyone with the token can control your bot. If it leaks, click Reset Token again and the old token stops working.

3 Add the Bot to Your Server and Copy the Server ID

  1. In the Developer Portal, open the Installation page and keep Guild Install selected.
  2. Add the bot scope and give it at least Send Messages and Read Message History.
  3. Copy the install link, open it in your browser, choose Add to server and pick your server.
  4. In Discord, turn on Developer Mode (User Settings › Advanced), right-click your server icon and choose Copy Server ID.

4 Choose Your AI Model

In OpenClaw Easy, open AI Provider in the sidebar and choose one:

  • Pre-paid credits: managed AI with no API key, pay as you go.
  • Bring Your Own Key: paste your Claude, OpenAI or Gemini API key.
  • Local LLM: connect to Ollama for private, offline AI (see the Ollama setup guide).
AI Provider settings in OpenClaw Easy

5 Connect Discord in OpenClaw Easy

In the sidebar, go to Channels and click Discord. Paste the bot token from step 2 and the Server ID from step 3, then click Connect. The app checks the token with Discord and registers your server.

OpenClaw Easy built-in AI chat interface

6 Test Your AI Discord Bot

In a server channel, type @YourBotName followed by a question. In server channels the bot replies when it is mentioned, so it will not answer every message. You can also send it a direct message, which needs no mention.

Customize behavior in Agent Config: change the model, edit the system prompt, and set fallbacks.

Popular Use Cases for AI Bots on Discord

Gaming Communities

An AI that answers game questions, explains mechanics, suggests builds, or writes lore. Put the game's rules and terminology in the system prompt so answers stay on topic.

Study Groups and Homework Help

Members mention the bot to explain a concept, check a proof, proofread an essay or quiz them before an exam, right in the study channel.

Developer Communities

A bot that explains error messages, suggests fixes and answers programming questions, useful in open-source servers where maintainers cannot answer every question.

Creative Writing and Roleplay

Collaborative storytelling, character creation and worldbuilding, with a persona defined in the system prompt. The model provider's usage policies still apply.

Community Support

If you run a product with a Discord community, the bot can answer frequent questions and point members to documentation before a human steps in.

Choosing the Right AI Model for Your Discord Bot

Claude (Anthropic)

Strong at long, nuanced conversations and following detailed instructions. A good fit for servers where discussions go deep.

GPT (OpenAI)

Broad general knowledge and a large ecosystem. A solid default for most servers. See GPT vs Gemini for a Discord bot.

Gemini (Google)

Strong multilingual support, which suits international communities.

Local Models (Ollama)

Maximum privacy, because data never leaves your machine. Use open models such as Llama, Qwen, Mistral or DeepSeek. Needs decent hardware (16 GB+ RAM recommended).

Tip: OpenClaw supports model fallback chains. Set a primary model with a cheaper fallback in case the primary is unavailable. Configure it in Agent Config.

Customizing Your Discord AI Bot

  • System prompt: define the bot's personality, knowledge domain and tone. Example: "You are a helpful assistant for the Rust programming community. Keep answers short and include code examples."
  • Mention gating: server messages require a mention by default. Turn that off per server or per channel with the requireMention setting in OpenClaw's Discord configuration if you want the bot to join every conversation in a dedicated channel.
  • Channel allowlists: restrict the bot to specific channels in OpenClaw's Discord guild settings.
  • Scheduled messages: use cron jobs to post a daily challenge or weekly recap.
  • Skills: add extra capabilities such as web search or GitHub integration.

Frequently Asked Questions

What is the best AI bot for Discord?

It depends on what you need. If your server already uses a community-management bot such as MEE6, its AI plugin is the least effort, but it needs a paid AI subscription and you cannot pick the model. If you want to choose the model (Claude, GPT, Gemini) and control the system prompt, run your own bot with your own API key. If privacy matters most, run your own bot on a local model through Ollama. OpenClaw Easy is a free desktop app that sets up the last two options without code.

Is there a free AI Discord bot?

Yes. OpenClaw Easy is a free desktop app for macOS and Windows that connects a Discord bot to an AI model. With a local model through Ollama, the whole setup costs nothing beyond your own computer's electricity. With a Claude, OpenAI or Gemini API key, the app stays free and you pay the AI provider for the tokens your server uses. Hosted bots with AI features usually put AI behind a paid tier.

Does Discord have a built-in AI bot?

No. Discord tested an experimental AI chatbot called Clyde, built with OpenAI technology, in 2023 and shut it down on December 1, 2023. To use AI in a Discord server today, you add a third-party bot or run your own bot connected to an AI model.

Can an AI bot read every message in my Discord server?

Only if it has the Message Content intent and can see the channel. Message Content is a privileged intent that the bot owner turns on in the Discord Developer Portal; apps whose servers reach more than 10,000 unique users must apply to Discord for it and reapply each year. A self-run OpenClaw Easy bot replies in server channels only when someone mentions it, unless you change that setting.

Does the bot work in DMs and group DMs?

An AI Discord bot set up with OpenClaw Easy answers direct messages and server channels. Group DMs are ignored by default in OpenClaw; they can be enabled for specific group DM channels in the OpenClaw Discord channel settings.

Does the bot need to be mentioned to respond?

In server channels, yes by default: OpenClaw only replies to guild messages that mention the bot, which keeps it from answering every message. Direct messages do not need a mention. You can turn off the mention requirement per server or per channel with the requireMention setting in OpenClaw's Discord configuration.

How much does it cost?

OpenClaw Easy is free and has no subscription. With your own Claude, OpenAI or Gemini API key you pay that provider directly, and the bill depends on the model and how many messages the bot answers. Local Ollama models cost nothing to run. There is also an optional pay-as-you-go pre-paid credit tier for managed AI access without an API key.

Will the bot run when my computer is off?

No. An OpenClaw Easy Discord bot runs on your own computer, so it only replies while the app is running. Minimize the app to the system tray to keep it running in the background, or run it on a machine that stays on if your server needs round-the-clock replies.

Is it safe to paste my Discord bot token into OpenClaw Easy?

The token is saved in the OpenClaw configuration file on your own machine and is not uploaded to any OpenClaw server; the bot connects from your computer directly to Discord. Treat the token like a password and do not share that file. If a token is ever exposed, click Reset Token on the Bot page of the Discord Developer Portal and the old token stops working.

What is Next?

An AI Discord bot you run yourself takes about ten minutes with OpenClaw Easy: no coding, no rented server. Download the free app and give your server an AI assistant on the model you choose.