Getting Started
Overview
Some of the most brilliant people and impactful projects in the world are hindered by bad prompts. Too many companies are integrating LLMs into production systems without any meaningful insights into what they're actually doing.
We became comfortable, accustomed to simply telling computers what to do. Instead we now find ourselves in tedious negotiations with our mechanical counterparts- cajoling, coercing, and manipulating an LLM, with maddeningly mixed results as models change or better models become available.
We became comfortable, accustomed to simply telling computers what to do. Instead we now find ourselves in tedious negotiations with our mechanical counterparts- cajoling, coercing, and manipulating an LLM, with maddeningly mixed results as models change or better models become available.
Libretto is here to fix that.
Quickstart
Quick guide to integrating Libretto and how to monitor your prompts in production.
Prompt Engineering Quickstart
Explore how Libretto can help you get the most out of your prompts.
Working With Prompt Templates
Take a deep dive in how to navigate and use Libretto.
SDK Integration
Learn about the Libretto SDK and how easily integrate into your existing LLM infrastructure.
What Libretto Can Do For You
At Libretto, we're tackling what we believe to be the most fascinating problem in AI: unlocking consistently great results from generative AI, transforming it from a novelty to a game-changer from something a few people can harness to a resource for all.
Our mission is clear: To make prompt monitoring and engineering fast, automated, fully empirical. Our tool is designed to test, improve & monitor LLM prompts rapidly, removing the guesswork & tedium.
Monitoring: Learning from your users
Libretto monitors your LLM calls in production, learning from your prompts and traffic to create a robust set of test cases and evals. We will automatically run these evals against a sampled portion of your calls, along with flagging any call deemed unsafe or toxic.
Testing: Finding feasibility, fast
Testing can confirm that an LLM is able to solve your problem — but it's plagued by trial & error. Libretto simplifies the crucial step of creating and managing test cases easy. We make it easy for users to tweak prompts, run tests, & get empirical results to separate the signal from the noise.
Improving: Building a real product through Experiments
"Experiments" is Libretto's killer prompt engineering tool, automatically exploring dozens or even hundreds of prompt variations to find the top performer. Imagine boosting prompt accuracy by 10% in minutes without guesswork. Goodbye to the days of painstaking manual testing!