Getting Started

TuringPulse is the control plane for AI agents — observe behavior, govern decisions, and intervene when it matters.

ℹ️
Cloud Platform
TuringPulse is available as a managed cloud platform. Sign up at app.turingpulse.ai to get your API key and start monitoring your AI agents.

Quick Start

Get started in under 5 minutes. Install the SDK, initialize with your API key, and start capturing traces.

1. Install the SDK

Installation
# Python
pip install turingpulse_sdk

# TypeScript/Node.js
npm install @turingpulse/sdk

2. Initialize & Instrument

agent.py
from turingpulse_sdk import init, instrument

# Initialize with your API key (get one at app.turingpulse.ai)
init(
    api_key="sk_live_your_api_key",
    workflow_name="my-project",
)

# Instrument your agent function
@instrument(name="my-agent")
def my_agent(query: str):
    response = llm.chat(query)
    return response

# Traces are automatically captured and sent to TuringPulse
result = my_agent("Hello, world!")

SDK Architecture

TuringPulse uses a plugin architecture — a lightweight core SDK with dedicated plugins for every framework and LLM provider. Install only what you need, or grab everything with one command.

Terminal
# Install everything
pip install turingpulse-sdk[all]

# Or pick specific plugins
pip install turingpulse-sdk turingpulse-sdk-openai turingpulse-sdk-langgraph

Framework Plugins

Auto-instrument your agent framework with zero code changes to your business logic. Each plugin produces standardized telemetry that flows to the TuringPulse platform.

LLM Providers

Intercept and trace LLM API calls with full token, cost, and latency capture. Works with both patch (global) and proxy (per-client) patterns.

What You Can Do

Once instrumented, TuringPulse automatically captures every agent execution and provides:

  • Full Traceability — Every decision traced back to inputs and steps
  • KPI Monitoring — Track latency, cost, tokens, and custom metrics
  • Drift Detection — Detect when agent behavior changes unexpectedly
  • Human Oversight — Route high-risk actions for human review
  • Alerts — Get notified when thresholds are breached

Next Steps