Welcome to the Mpalo API Overview

1. Private Data Spaces for Your Users

(Available on Pro and Enterprise plans. Compare all plans ›)

For businesses and applications serving multiple end-users, Palo introduces Private Data Spaces (Beta). This feature provides a scalable, managed solution for deploying AI memory to your customers, without the burden of managing individual vector stores for each one if you choose not to use the BYOVS model for this purpose. This provides a powerful alternative to the BYOVS model for businesses that prefer a fully managed, scalable data environment provided directly by Mpalo.

Within your dedicated Private Data Space, Mpalo securely shards and isolates each of your end-user's memories, ensuring their data remains strictly private and accessible only to them and your application under their permission. This means you can onboard thousands or millions of your customers, each with their own secure, personalized memory, all managed seamlessly by Mpalo. This feature is designed to simplify scaling your AI applications for multi-user environments while upholding stringent data privacy and is configurable in the Workbench for Pro and Enterprise tiers.

2. Bring Your Knowledge: Document-to-Memory Pipeline

(Available on Pro and Enterprise plans. Compare all plans ›)

Transform your raw documents – from PDFs and text files to CSVs and spreadsheets – into rich, queryable memories that Palo can learn from and use. Our new Document-to-Memory Pipeline (Beta) makes it easy to integrate your existing knowledge bases directly into Palo's memory system.

This pipeline helps you build powerful Retrieval-Augmented Generation (RAG) applications and create custom, data-infused conversational experiences. Your original document content is processed for embedding (using a format string you define for optimal semantic richness), but the raw document itself is not permanently stored by Mpalo unless you explicitly choose a storage option. We generate the necessary vectors from your processed content, which are then placed into your designated private vector store (either Mpalo-managed or BYOVS), ensuring your underlying data remains yours and processed according to your instructions. Access this feature via the Workbench (Pro/Enterprise).

This pipeline is the first step in creating a powerful custom memory: process your files with the Document Pipeline, store the vectors in your BYOVS or a Private Data Space, and then use a Memory Template to define how Palo interacts with that knowledge. Explore how to best structure your data for Palo and the detailed ingestion flow in our Memory Templates documentation.