a1facts is a powerful Python framework that acts as a verifiable knowledge layer for factual, precise AI. It allows you to define a clear structure (ontology) for your domain, ingest verifiable information from various sources, and use natural language to get precise, reliable answers. By grounding your AI agents in a structured knowledge graph, a1facts helps eliminate hallucinations and ensures that your agent’s responses are based on verified facts. This project supports a Neo4j backend to manage and query the graph data, providing a robust and scalable solution for your knowledge management needs.
a1facts Architecture Diagram

​
Key Features

Ontology-Driven

Define your knowledge domain with simple YAML files. a1facts uses this ontology to automatically create the tools and structures needed to interact with your data.

Verifiable Ingestion

Convert unstructured knowledge into structured entities and relationships within your knowledge graph, ensuring data quality and consistency.

Natural Language Queries

Ask questions in plain English. a1facts translates your queries into precise graph traversals to retrieve the information you need.

Reliable AI Agents

Build AI agents that can provide accurate and trustworthy answers by grounding them in the factual data of the knowledge graph.

​
How a1facts Works

In many applications, it’s critical for AI agents to provide answers that are not just plausible but also precise and factually correct. Standard large language models (LLMs) can sometimes “hallucinate” or generate incorrect information, which can be a major issue in domains like finance, legal, or scientific research. a1facts addresses this problem by grounding your AI agent in a knowledge graph. Here’s how it works:
1

Define Your Domain

You start by creating an ontology that describes the key concepts and relationships in your specific domain. For example, in finance, you might define entities like Company, Product, and Market, and relationships like competes_with or operates_in.
Example of an ontology definition in YAML
2

Populate the Graph

You then ingest data from different sources into the knowledge graph. This could be from internal documents, databases, external APIs, or the web. a1facts ensures that this data is structured according to your ontology, but also that it’s factual by triangulating facts from different sources.
Visualization of a knowledge graph with entities and relationships
3

Query with Confidence

When your AI agent receives a question, it doesn’t just rely on its internal training data. Instead, it uses a1facts to query the knowledge graph. This means the agent’s answers are based on the structured, verified data.
Example of a natural language query and the corresponding graph traversal
By using a1facts, you can build an AI agent that is not only intelligent but also trustworthy, providing precise and reliable answers every time.
a1facts is a work in progress. We’d love to hear from you on how you’re using it and what you need it to do next.