Papr Memory API

The universal memory layer that lets AI apps remember, reason, and grow overnight.

TL;DR: Store with POST /v1/memory. Retrieve with POST /v1/memory/search.
Our purpose-built vector+graph architecture with predictive intelligence gives you multi-hop reasoning that delivers 46% higher accuracy than OpenAI.

Key Features

• Real-time content ingestion – Text, chat, PDFs/Docs via SDK or API
• Multi-hop reasoning - Not just vector+graph bolted together; designed from the ground up for scalable graph traversal.
• Dynamic indexing and re-ranking - memories are dynamically indexed overtime and re-ranked for the best results.
• Custom domain schemas – Define multiple ontologies once, to guide graph extraction.
• Perosnaliztion + self-improvement – Support for both user memories (personalization) and agent memories (self-improvement).
• Natural language + GraphQL – Use natural language search to retreive context or GraphQL to analzye context and get insights.
• Predictive context caching – Predictive models anticipate what you'll need and cache context in advance for lightning-fast retrieval.
• Granular ACL & sharing – User-level filters executed inside the engine with permissions for users and agents.

How It Works

Two main operations. POST /v1/memory to add memories. POST /v1/memory/search to retrieve with both semantic, relational recall, and re-ranking. GraphQL for analyzing and getting insights on stored memories and context.

Under the hood, Papr optimally chunks, stores, and connects every memory dynamically:

Three Input Paths

1. Documents (POST /v1/document) - Upload PDFs or Word docs. System analyzes and selectively creates memories.
2. Messages/Chat - Send conversation history. System analyzes and extracts important information.
3. Direct Memory (POST /v1/memory) - Explicitly create memories with full control. Perfect for agent self-documentation.

Memory Engine Intelligence

• Vector embeddings for semantic similarity search
• Knowledge graphs automatically connect memories and build knowledge graph
• Predictive caching anticipates context needs for fast retrieval

Two Query Modes

1. Natural Language Search (POST /v1/memory/search) - Ask questions, get relevant & re-ranked memories + graph entities. Semantic + graph combined.
2. GraphQL (POST /v1/graphql) - Run structured queries for analytics, aggregations, and relationship analysis.

What You Can Build

Dual Memory Types

Papr supports two types of memories, enabling comprehensive AI capabilities:

• User Memories - Information about users: preferences, history, context, conversations. Enables personalization.
• Agent Memories - Agent documents its own workflows, learnings, reasoning patterns. Enables self-improvement.

Both stored and queried the same way, allowing agents to not just personalize for users, but to learn and improve their own capabilities over time.

Next Steps