Skip to main content

Simple AI Chat Template

The simplest possible Klyntos workflow - a chat interface that responds using an AI model.

Time to build: 5 minutes
Difficulty: Beginner
Perfect for: Getting started, testing models, quick prototypes

What It Does​

User sends a message → AI responds with an answer

That's it! Simple, but powerful foundation for more complex workflows.

Step-by-Step​

1. Create Workflow​

  • New workflow → Name it "Simple Chat"
  • Default Start Block is already there

2. Add Agent Block​

  • Drag Agent Block onto canvas
  • Connect Start block output to Agent input

3. Configure Agent​

Model: GPT-4o (fast and cheap)

System Prompt:
"You are a helpful AI assistant. Provide clear, concise answers."

User Prompt: <start.input>

4. Add Response​

  • Drag Response Block
  • Connect Agent to Response
  • Content: <agent.content>

5. Test​

  • Open Chat panel
  • Type "Hello!"
  • See AI response

Done! 🎉​

You've built your first workflow. Now enhance it:

Make it smarter​

  • Add Knowledge Base tool for context
  • Add Thinking tool for complex reasoning
  • Use Claude 3.7 for better responses

Make it useful​

  • Add Condition to route different question types
  • Add Memory to remember conversation history
  • Add Evaluator to check response quality

Deploy it​

Cost​

Per 1,000 messages:

  • GPT-4o: ~$2-3
  • Claude 3.7 Sonnet: ~$10-15
  • GPT-4o-mini: ~$0.20-0.40
Next Step

Try the Customer Support Agent to see how to add knowledge bases and escalation logic!