Ultra-low latency knowledge graph API and SDK for voice AI applications
Start for Free
Ingest from PDF, Word and Text documents
Ingest a single web page or an entire website
Ingest a video file with one or more speakers
Ingest an audio file with one or more speakers
Serve memory quickly with precision
Serve large memory sets quickest
pip install duohubfrom duohub import Duohub
client = Duohub()
result = client.query("Where is Ryan going in two weeks?")What is a knowledge graph?
A knowledge graph is a structured database that represents information as a network of connected entities and their relationships - similar to how humans naturally think about and connect ideas. Knowledge graphs are particularly powerful because they can capture complex, real-world relationships in a way that's both machine-readable and logically structured, making them ideal for AI applications that need to understand context and make connections between different pieces of information.
What types of content can I add to my knowledge graph?
You can add content from various sources:
How does the content processing work?
Content processing involves several steps:
The initial preprocessing step is free.
What is ingestion and how is it priced?
Ingestion is the process of uploading and processing your data to create your knowledge graph. This includes analyzing text, audio, or other forms of data to understand your communication style and knowledge base. Ingestion is priced per 1,000 tokens. This aligns with our processing costs.
Why is graph RAG better than vector RAG?
The key advantage of graph RAG over vector RAG is that it can return related content that is not semantically similar, whereas vector RAG can only return content that is semantically similar. Graph RAG is more accurate than vector RAG in all benchmarks.
Can I bring my own ontology?
Yes, we have a range of ontologies that you can choose from. You can also bring your own ontology, or we can work with you to develop an ontology for your use case.
