Quickstart

This guide gets you from zero to seeing LLM traces in your Promptic dashboard.

1. Install the SDK

pip install promptic-sdk[openai]

Replace [openai] with the extra that matches your stack (e.g. [anthropic], [bedrock], [vertexai], [mistralai], [langchain], [openai-agents], [claude-agent]), or use [all] to install every supported instrumentor. See Installation for the full list.

2. Get your API key

Go to your workspace settings and create a new API key. Set it as an environment variable:

export PROMPTIC_API_KEY="pk_live_..."

3. Add two lines to your code

import promptic_sdk

promptic_sdk.init()

Call init() once at startup, before any LLM calls. That's it — all LLM calls are now automatically traced.

4. Tag traces with a component

Wrap your LLM calls in an ai_component context to organize traces:

import promptic_sdk
from openai import OpenAI

promptic_sdk.init()
client = OpenAI()

with promptic_sdk.ai_component("my-classifier"):
    response = client.chat.completions.create(
        model="gpt-4.1-nano",
        messages=[{"role": "user", "content": "Is this email spam? ..."}],
    )
    print(response.choices[0].message.content)

5. View your traces

Open the Promptic dashboard. Your traces appear under the component you specified, showing the full request/response, token counts, cost, and latency.

What's next?

Now that tracing is set up, explore what you can do:

Tracing guide — Datasets, runs, multi-provider setup, and advanced configuration
Prompt Optimization — Automatically find the best prompt for your task
Agent Evaluation — Evaluate your AI agents with structured test datasets
Deployment — Deploy optimized prompts and fetch them at runtime