How to Install Flowise with Docker: AI Agent Builder Setup Guide

Quick Install with Docker

The fastest way to run Flowise is a single Docker command. This starts Flowise on port 3001 with an in-memory SQLite database.

docker run -d \
  -p 3001:3001 \
  --name flowise \
  -v ~/.flowise:/root/.flowise \
  flowiseai/flowise

What each flag does:

• `-d` — runs the container in the background (detached mode)
• `-p 3001:3001` — maps Flowise's port to your machine
• `-v ~/.flowise:/root/.flowise` — persists your flows, credentials, and chat history to your home directory
• `flowiseai/flowise` — the official Docker Hub image

Wait 30 seconds for the container to start, then open your browser at:

http://localhost:3001

You should see the Flowise canvas with an empty chatflows list.

Add Basic Authentication

By default Flowise has no login protection. Anyone who can reach port 3001 can access and modify your flows. Add a username and password with environment variables:

docker run -d \
  -p 3001:3001 \
  --name flowise \
  -v ~/.flowise:/root/.flowise \
  -e FLOWISE_USERNAME=admin \
  -e FLOWISE_PASSWORD=changeme123 \
  flowiseai/flowise

Docker Compose Setup with Persistent Storage

Docker Compose is the recommended approach for a stable Flowise deployment. It gives you persistent SQLite storage, easy environment variable management, and a configuration file you can version-control.

Create a project directory:

mkdir -p ~/flowise && cd ~/flowise
nano docker-compose.yml

Paste the following compose file:

version: '3.8'

services:
  flowise:
    image: flowiseai/flowise
    container_name: flowise
    restart: unless-stopped
    ports:
      - "3001:3001"
    volumes:
      - flowise_data:/root/.flowise
    environment:
      - PORT=3001
      - FLOWISE_USERNAME=admin
      - FLOWISE_PASSWORD=strongpasswordhere
      - DATABASE_PATH=/root/.flowise
      - SECRETKEY_PATH=/root/.flowise
      - LOG_PATH=/root/.flowise/logs
      - BLOB_STORAGE_PATH=/root/.flowise/storage
      # Uncomment to connect to a local Ollama instance
      # - OLLAMA_BASE_URL=http://host.docker.internal:11434

volumes:
  flowise_data:

Start Flowise:

docker compose up -d

Check it is running:

docker compose logs -f flowise

You should see output similar to:

flowise  | Ready on port 3001

Stop with `docker compose down`. Restart with `docker compose up -d`. The `flowise_data` volume persists your flows and credentials across container restarts and image upgrades.

Connect Flowise to Local Ollama Models

Flowise connects to Ollama through a dedicated ChatOllama or OllamaEmbeddings node. This lets you build AI agents and RAG pipelines that run entirely offline with no API costs.

Step 1: Ensure Ollama is Accessible

If Ollama is running natively on the same host as Flowise, the container reaches it via the `host.docker.internal` DNS name:

# Test from inside the Flowise container
docker exec flowise curl -s http://host.docker.internal:11434
# Expected: Ollama is running

If Ollama is in its own Docker container, use the container's service name instead of `host.docker.internal`. Add Ollama to your compose file and reference it as `http://ollama:11434`.

Step 2: Add a ChatOllama Node in Flowise

1. Open Flowise at `http://localhost:3001` 2. Click **Add New** to create a chatflow 3. In the nodes panel (left sidebar), search for **ChatOllama** 4. Drag the ChatOllama node onto the canvas 5. In the node configuration: - **Base URL**: `http://host.docker.internal:11434` - **Model Name**: type `llama3.3` or any model you have pulled with `ollama list`

Step 3: Build a Simple Chain

Connect the ChatOllama node to a **Chat Prompt Template** node and a **Conversation Chain** node:

[Chat Prompt Template] → [ChatOllama] → [Conversation Chain]

Click **Save** at the top right, then **Upsert** to activate the flow. Open the chatbot preview to test a conversation.

Build Your First AI Agent

Flowise's agent nodes use LangChain's agent framework to let your LLM call tools, search the web, query databases, or run code. This example builds a simple ReAct agent with web search capability.

Step 1: Create a New Agentflow

Click **Agentflows** in the top navigation (not Chatflows). Agentflows are designed specifically for agent architectures with tool use.

Step 2: Add the Required Nodes

Connect the nodes:

[Start] → [OpenAI Tool Agent] → output
[ChatOpenAI] → [OpenAI Tool Agent] (LLM input)
[Calculator] → [OpenAI Tool Agent] (tool)
[SerpAPI] → [OpenAI Tool Agent] (tool)

Step 3: Test the Agent

Save and open the chatbot preview. Ask a question that requires web search:

What is the current price of Nvidia stock and what is 523 * 47?

The agent should use SerpAPI for the stock price and Calculator for the math. You will see the tool calls logged in the chat panel on the right.

Expose Flowise as an API or Embed a Chatbot

Every Flowise chatflow automatically generates an API endpoint and an embeddable chatbot widget. This is how you deploy Flowise-built chatbots on your own website or call them from your application.

API Endpoint

After saving a chatflow, click the **API Endpoint** button (top right, looks like ``). Flowise shows:

# Example API call to your Flowise chatflow
curl http://localhost:3001/api/v1/prediction/YOUR_CHATFLOW_ID \
  -H "Content-Type: application/json" \
  -d '{"question": "What is Flowise?"}'

Replace `YOUR_CHATFLOW_ID` with the UUID shown in the API endpoint panel. Add an API key for authentication:

# With API key authentication
curl http://localhost:3001/api/v1/prediction/YOUR_CHATFLOW_ID \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -d '{"question": "What is Flowise?"}'

Generate API keys in Flowise under **Settings > API Keys**.

Embeddable Chat Widget

Click **Embed** (next to API Endpoint). Flowise provides a JavaScript snippet you paste into any HTML page:

<script type="module">
  import Chatbot from "https://cdn.jsdelivr.net/npm/flowise-embed/dist/web.js"
  Chatbot.init({
    chatflowid: "YOUR_CHATFLOW_ID",
    apiHost: "https://your-flowise-domain.com",
  })
</script>

The widget renders a floating chat button in the bottom-right corner. It connects directly to your Flowise instance, so keep Flowise running and publicly accessible for the widget to work.

Troubleshooting

Frequently Asked Questions