Quick Start

Get started with Dakora in 5 minutes. Create your API key, use starter templates, and render your first prompt.

Follow these steps to start optimizing your AI costs with Dakora.

1. Create your account

   Sign up at Dakora Studio to get started.

   Each project has its own templates, API keys, and cost analytics. It perfect for separating development, staging, and production environments.

2. Generate an API key

   Navigate to Settings → API Keys and click New Key. Give your key a name, set an expiration (or choose “Never”), and click Create Key.

   

   Copy your API key immediately - it’s only shown once. Store it securely.

3. Explore starter templates

   Every new project comes with starter templates to help you get going quickly.

   

   You can use these templates as-is, customize them, or create your own from scratch in the Studio.

4. Render and execute your first prompt

   Install the Dakora SDK and instrumentation packages:

   pip install dakora dakora-instrumentation
   pip install opentelemetry-sdk opentelemetry-exporter-otlp-proto-http
   pip install openai opentelemetry-instrumentation-openai
   

   Make your API keys available as environment variables: DAKORA_API_KEY and OPENAI_API_KEY (via .env file or your preferred method).

   Now set up instrumentation to track your LLM calls, then render a template and use it with your LLM:

   import asyncio
   import os
   from opentelemetry.instrumentation.openai import OpenAIInstrumentor
   
   # Instrument OpenAI before importing it
   OpenAIInstrumentor().instrument()
   
   from dakora import Dakora
   from dakora_instrumentation.generic import setup_instrumentation
   from openai import OpenAI
   
   async def main():
       # Initialize Dakora client
       dakora = Dakora()
   
       # Setup instrumentation to track executions
       setup_instrumentation(
           dakora_client=dakora,
           service_name="my-app"
       )
   
       # Render the FAQ responder template
       result = await dakora.prompts.render(
           "faq_responder",
           {
               "question": "How do I reset my password?",
               "knowledge_base": "Users can reset passwords via Settings > Security > Reset Password.",
           },
       )
   
       # Use the rendered template with OpenAI (automatically traced)
       client = OpenAI(api_key=os.getenv("OPENAI_API_KEY"))
       response = client.chat.completions.create(
           model="gpt-4o-mini",
           messages=[{"role": "user", "content": result.text}],
       )
   
       print(response.choices[0].message.content)
   
       await dakora.close()
   
   asyncio.run(main())
   

   The instrumentation automatically captures token usage, costs, and latency for every LLM call.

5. View your execution

   Check the Dakora Studio executions to see your trace with full analytics.

   

   See token usage, costs per model, and identify optimization opportunities.

Next Steps

Templates

Learn how to create and manage prompt templates

Try the Studio

Test your templates with real inputs before deploying

Cost Analytics

Understand your execution logs and cost data

Budget Controls

Set up spending limits and alerts