Welcome to Inductor#

Inductor is a developer tool for evaluating, monitoring, and improving your LLM applications – both during development and in production.

Why Inductor#

Inductor makes it easy to continuously answer questions like

  • How good is our LLM application?

  • How can we improve its quality and cost-effectiveness?

  • Are we actually improving?

  • Is our LLM app production-ready?

  • How is our LLM app behaving on live traffic?

so that you can deliver high-quality, cost-effective, and safe LLM apps far more easily and quickly.

One-minute integration#

  • Start running Inductor on your LLM app with zero code modifications.

  • Works seamlessly with your existing stack (any model, any way of writing LLM apps).

  • Runs in your existing environment. Stay in the code editor or notebook environment that you already use.

Key features#

Inductor provides a powerful set of capabilities for developing fast, shipping reliably, and collaborating easily:

  • Test your LLM app easily, rigorously, and continuously using Inductor’s test suites and CLI.

  • Stay organized effortlessly with automatic tracking and versioning of everything, execution auto-logging, human + machine evaluations, and Inductor’s web UI.

  • Turbocharge your experimentation with Inductor’s hyperparameters.

  • Rapidly assess your LLM app and determine how to improve quality and cost-effectiveness using Inductor’s purpose-built analytical tools.

  • Understand usage, detect issues, and seamlessly feed back into your development process using Inductor’s rich, automated production logging.

Get started#

Go to the Quickstart guide to learn how to use Inductor.