Getting Started

Quick Start

Vyrt gives you full observability into your AI agents — every run, step, tool call, and failure captured automatically. This guide gets you instrumented and sending data to the dashboard in under five minutes.

Install the SDK

Add the Vyrt SDK to your project using your preferred package manager.

1
pip install vyrt

Initialize Vyrt

Import the SDK and configure your API key. Find your key in Settings → API Keys.

1
2
3
from vyrt import create_client

vyrt = create_client(api_key="vk_your_api_key_here")

Instrument Your Agent

Start a run, record steps as your agent executes, then close the run. Every invocation is captured end-to-end.

1
2
3
4
5
6
7
8
9
10
11
12
def run_my_agent(user_input: str):
    run = vyrt.start_run(name="customer-support-agent", model="claude-3-5-sonnet")
    try:
        run.add_step(name="lookup", input=user_input, output="Found answer", duration=320)
        run.end()
        return "Done"
    except Exception as e:
        run.fail(str(e))
        raise

if __name__ == "__main__":
    run_my_agent("How do I reset my password?")

Verify in the Dashboard

Trigger a single run, then navigate to Agent Runs — your trace should appear within a few seconds.

bash
1
2
3
4
5
6
# Trigger a single run to generate trace data
python agent.py

# Open the Vyrt dashboard and navigate to:
#   Dashboard → Agent Runs
# Your run will appear within a few seconds.

Product

ObservabilityInsightsIntegrations

Resources

DocsBlogChangelogSDK Reference

Company

AboutPricingSign InContact

Legal

Privacy PolicyTerms of Service

Full visibility into every AI decision in production.

© 2026 Vyrt. All rights reserved.