Langflow
Getting Started

Getting Started

Build your first AI workflow in minutes.

Launching Langflow

  1. Open the Hub
  2. Click Langflow in the tool list
  3. Wait for the environment to load
  4. You’ll see the visual editor

Interface Overview

Canvas

The main workspace where you build flows:

  • Drag components here
  • Connect them with lines
  • Arrange your workflow visually

Component Sidebar

Left panel with available components:

  • Inputs: Chat, Text, File
  • LLMs: OpenAI, Anthropic, Ollama
  • Prompts: Templates, System Messages
  • Memory: Conversation storage
  • Tools: Search, Calculator, Code
  • Outputs: Chat, Text, JSON

Properties Panel

Right panel showing:

  • Selected component settings
  • Parameter configuration
  • Connection details

Toolbar

Top bar with:

  • Save/Load flows
  • Run/Test buttons
  • Settings access
  • Export options

Your First Flow: Simple Chatbot

Step 1: Add Input

  1. Find Chat Input in the sidebar
  2. Drag it onto the canvas
  3. This receives user messages

Step 2: Add LLM

  1. Find OpenAI (or your preferred LLM)
  2. Drag it onto the canvas
  3. Position it to the right of the input

Step 3: Add Output

  1. Find Chat Output in the sidebar
  2. Drag it onto the canvas
  3. Position it to the right of the LLM

Step 4: Connect Components

  1. Click the output port (right side) of Chat Input
  2. Drag to the input port (left side) of OpenAI
  3. Click OpenAI’s output port
  4. Drag to Chat Output’s input port

You should see:

[Chat Input] → [OpenAI] → [Chat Output]

Step 5: Configure LLM

  1. Click on the OpenAI component
  2. In the properties panel:
    • Select model: gpt-4o or gpt-3.5-turbo
    • API key: Enter yours or use workspace default
  3. Save settings

Step 6: Test Your Flow

  1. Click Run in the toolbar
  2. The Playground opens
  3. Type “Hello!” and press Enter
  4. See your chatbot respond!

Adding a System Prompt

Make your chatbot have a personality:

Step 1: Add Prompt Component

  1. Find Prompt in the sidebar
  2. Drag it between Input and LLM

Step 2: Configure the Prompt

  1. Click the Prompt component
  2. Enter your system message:
You are a helpful assistant that specializes in cooking.
You always suggest recipes and cooking tips.
Keep responses concise and practical.

Step 3: Rewire Connections

  1. Delete old Input → LLM connection
  2. Connect: Input → Prompt → LLM → Output

Step 4: Test Again

Now your chatbot will respond as a cooking assistant!

Adding Memory

Make your chatbot remember the conversation:

Step 1: Add Memory Component

  1. Find Conversation Buffer Memory
  2. Drag it onto the canvas

Step 2: Connect Memory

Connect memory to your LLM component:

  1. Memory output → LLM memory input
  2. This gives the LLM conversation context

Step 3: Configure Memory

Set how much history to keep:

  • k: Number of exchanges to remember
  • Typical: 5-10 exchanges

Step 4: Test Conversation

  1. Ask: “What’s a good pasta dish?”
  2. Then ask: “How long does it take to make?”
  3. The bot remembers you’re discussing pasta!

Saving Your Flow

Save to Workspace

  1. Click Save in toolbar
  2. Name your flow: “Cooking Chatbot”
  3. Click Save

Your flow is now saved in your Calliope workspace.

Export as JSON

  1. Click Export in toolbar
  2. Choose JSON
  3. Download the file

Use this to:

  • Version control in Git
  • Share with teammates
  • Import into other environments

Next Steps

Now that you’ve built a basic chatbot:

Components