​

Pongo is the one line of code to improve and monitor your RAG performance.

Our filter usees a collection of models to improve search accuracy at runtime, and our observe uses larger models to build detailed accuracy reports in the background

​

How it works

1. Follow the quick start to get your line of code and api keys
2. Retrieve the top ~200 results from your pipeline
3. Send the results to pongo’s semantic filter for a 35% improvement in search accuracy
4. Make sure the observe variable is True in your calls
5. We’ll analyze queries and email you a weekly report of your pipeline’s accuracy, and you can check your accuracy stats at any time in the analytics dashboard.
6. You can also set up custom alerts for when certain queries return with no relevant context

​

Start without changing your pipeline

If you want to get started without changing your pipeline’s behavior, check the standalone observability guide!

​

Start without observing your pipeline

If you just want higher accuracy without evaluating your query performance, check out the filter guide and set the observe flag to false.

​

Get Started