Build a Lark (Feishu) AI Bot with Ollama

This recipe puts a Ollama model behind a Lark (Feishu) bot. LangBot receives messages from Lark (Feishu), runs them through a pipeline, and sends the answer back, so you do not write any platform code.

Run open models locally with Ollama and serve them through LangBot's built-in Agent, so conversations never leave your machines.

What you need

• A running LangBot instance: LangBot Cloud (no server needed) or a self-hosted deployment with Docker or uvx langbot.
• A Ollama account or server you can call from LangBot.
• A Feishu or Lark custom app (app_id and app_secret). The one-click setup in LangBot can create and authorize it for you.

Networking

The default WebSocket long connection needs no public IP. Where long connection is unavailable (for example some Lark international setups), webhook mode requires a public HTTPS address; LangBot Cloud includes one.

1. Add the Ollama model

1. Install Ollama and pull a model (for example ollama pull followed by the model name).
2. In LangBot, open Models and add an LLM model with the provider set to Ollama and the request URL pointing at your Ollama server (by default port 11434).
3. If LangBot runs in Docker, localhost is the container itself: use the host's address (for example host.docker.internal) instead.

2. Create a pipeline

Open Pipelines in the LangBot WebUI, create a pipeline (or edit the default one) and set AI as follows.

3. Create the Lark (Feishu) bot

1. Create a bot with the Lark adapter, select the platform domain (Feishu or Lark) and use One-Click Setup, or enter app_id and app_secret manually.
2. Add the message permissions and event subscriptions from the guide (and the card permissions if you want streaming cards), then publish an app version.

4. Bind and test

Open the bot, select the pipeline you created, save and make sure the bot is enabled.

Search for the bot in Feishu/Lark, send it a direct message, then add it to a group and mention it.

Troubleshooting

• Small local models may not support tool calling well; leave Tool Use off for models that do not support it.
• No messages arrive: confirm the app version is published and the event subscription uses the same mode (long connection or webhook) as LangBot.
• Check Monitoring in the WebUI: every message, pipeline run and error is recorded there.

FAQ

Do I need a server with a public IP for Lark (Feishu)?

Not necessarily. The default WebSocket long connection needs no public IP. Where long connection is unavailable (for example some Lark international setups), webhook mode requires a public HTTPS address; LangBot Cloud includes one.

Does the Lark (Feishu) bot support streaming replies?

Yes. Enable streaming on the bot; the AI back end must also stream. See the platform feature matrix for details.

Can the same Ollama setup serve other platforms too?

Yes. A pipeline can be bound to many bots, so one configuration can answer on several platforms at once.

Related guides

• Full Lark (Feishu) setup guide
• Model configuration
• Other AI back ends on Lark (Feishu): Dify · n8n · Coze · Langflow · DeepSeek
• Ollama on other platforms: Telegram · Discord · Slack · DingTalk · WeCom (internal app) · WeCom Intelligent Bot