Workflow:FlowiseAI Flowise Chatbot Deployment
| Knowledge Sources | |
|---|---|
| Domains | Deployment, API_Integration, LLM_Ops |
| Last Updated | 2026-02-12 07:30 GMT |
Overview
End-to-end process for deploying a Flowise chatflow as a production-ready chatbot through public sharing, website embedding, API integration, and Docker containerization.
Description
This workflow covers the complete deployment pipeline for making a Flowise chatflow accessible to end users. Once a chatflow is built and tested, it can be deployed through multiple channels: a shareable public URL, an embeddable chat widget (HTML or React component), or a REST API endpoint with generated client code in Python, JavaScript, and cURL. The workflow includes configuring security controls (API keys, rate limiting, domain whitelisting), customizing the chatbot's visual appearance, and deploying the Flowise server itself via Docker with options for distributed queue-based processing.
Usage
Execute this workflow after you have a tested, working chatflow that needs to be made available to end users or integrated into an external application. This applies when you need to share the chatbot via a link, embed it in a website, integrate it programmatically via API, or deploy the entire Flowise platform to a production server.
Execution Steps
Step 1: Configure Chatflow Security
Open the chatflow configuration dialog from the canvas header. Navigate to the Security tab to set up access controls. Configure rate limiting by specifying the maximum number of messages per time window and a custom message shown when the limit is exceeded. Set up allowed domains to restrict which websites can embed the chatbot. Configure override settings to control which node parameters can be modified at runtime via the API.
Key considerations:
- Rate limiting prevents abuse by capping requests per duration (e.g., 5 messages per 60 seconds)
- Domain whitelisting restricts embed access to approved origins
- Override configuration lets API callers customize specific parameters at prediction time
- Additional configuration tabs cover starter prompts, follow-up prompts, speech-to-text, text-to-speech, chat feedback, lead capture, file upload, and post-processing
Step 2: Create API Keys
Navigate to the API Keys management page. Create a new API key by specifying a name and selecting permissions. The system generates a unique key that can be associated with specific chatflows. API keys enable authenticated access to the prediction endpoint and control what operations the key holder can perform.
Key considerations:
- Each API key has configurable permissions (chatflows, templates, apikeys, etc.)
- Keys can be associated with specific chatflows for scoped access
- The generated key should be stored securely as it cannot be retrieved after creation
- API key authentication is optional; chatflows can also be made fully public
Open the Share Chatbot panel from the canvas header or chatflow settings. Toggle the "Make Public" switch to enable unauthenticated access. Copy the generated public URL which follows the pattern: baseURL/chatbot/chatflowId. Anyone with this URL can access the chatbot through a full-page chat interface powered by the flowise-embed library.
Key considerations:
- Public chatbots are accessible without authentication
- If an API key is required on the chatflow, the public page shows an access error
- The public page renders a full-page chat using the flowise-embed-react library
- Public access can be toggled off at any time
Step 4: Customize Chatbot Appearance
In the Share Chatbot panel, configure the visual styling of the embedded chatbot. Customize the title (text, avatar, colors), general settings (welcome message, error message, background color, font size), bot message styling (background, text color, avatar), user message styling, and text input appearance (colors, placeholder text, send button color).
Key considerations:
- All visual settings are stored in the chatbotConfig JSON field on the chatflow
- Settings apply to both the public URL page and embedded widgets
- Custom welcome messages and starter prompts improve user onboarding
- Disclaimer and terms acceptance can be configured for compliance
Step 5: Embed in Website
Open the Embed Chat tab to get integration code. Four embed formats are available: popup HTML (floating bubble widget), fullpage HTML (full-screen chat), React popup component (BubbleChat), and React fullpage component (FullPageChat). Copy the generated code snippet and add it to the target website. The embed code references the flowise-embed library from CDN and connects to the Flowise API host.
Key considerations:
- HTML embeds use script tags that load the flowise-embed library from CDN
- React embeds use the flowise-embed-react npm package
- Popup mode creates a floating chat bubble; fullpage mode fills the container
- The chatflowid and apiHost parameters are required in all embed configurations
- Additional customization options are available through the embed API
Step 6: Integrate via API
Open the API Code dialog from the canvas header. Select the programming language (Python, JavaScript, or cURL) and optionally assign an API key for authentication. Copy the generated code which sends POST requests to the prediction endpoint at /api/v1/prediction/chatflowId. The code handles message sending, streaming responses, and file uploads.
Key considerations:
- Python code uses the requests library for HTTP calls
- JavaScript code uses the native fetch API
- cURL examples provide direct command-line integration
- Streaming prediction uses a separate endpoint for server-sent events
- Override parameters can be passed to customize node behavior per request
Step 7: Deploy with Docker
For production deployment, use the provided Docker Compose configurations. The basic deployment starts a single Flowise container with persistent storage. For high-traffic scenarios, the queue mode deployment adds Redis and worker containers for distributed processing. Configure environment variables for database connection, storage backend (local, S3, or GCS), JWT authentication, and other server settings.
Key considerations:
- Basic deployment: docker compose up -d with port 3000 exposed
- Queue mode adds Redis and BullMQ workers for distributed job processing
- Data persists at ~/.flowise on the host via volume mount
- Multiple database backends supported: SQLite, PostgreSQL, MySQL
- Storage options include local filesystem, AWS S3, and Google Cloud Storage
- Health checks verify the service is running via /api/v1/ping
- Environment variables control all configuration aspects