What is Perseus

What is Perseus?

Perseus is Lettria’s Text-to-Graph engine.
It transforms unstructured text—documents, articles, emails, reports—into a structured knowledge graph that AI systems can reliably understand, query, and reason over.

Modern AI is extremely powerful, but it still struggles with one fundamental problem:
most organizational knowledge lives in raw text, while AI works best with explicit, structured, and connected data.

Perseus bridges this gap.

The problem it solves

Unstructured text hides valuable information:

• Entities are implicit
• Relationships are scattered across sentences and documents
• Queries require brittle keyword search or expensive prompt engineering
• Reasoning across documents is unreliable

AI systems, on the other hand, need:

• Clearly identified entities
• Explicit relationships
• A consistent schema
• Data that can be queried, filtered, and traversed

Perseus converts text into connected knowledge, making it usable far beyond simple retrieval.

What it does

At a high level, Perseus:

1. Analyzes raw text
2. Extracts entities and relationships
3. Maps them to a structured schema
4. Outputs a knowledge graph ready for downstream use

This graph can then power:

• Semantic and hybrid search
• Retrieval-augmented generation (RAG)
• Multi-document reasoning
• Analytics and insights
• Agent memory and long-term knowledge

Ontology-driven by design

A core strength of Perseus is that it is ontology-driven.

Instead of producing a generic or opaque graph, Perseus allows you to:

• Define your own ontology (entities, relations, constraints)
• Control what gets extracted
• Enforce domain-specific structure
• Reduce noise and ambiguity

This makes Perseus especially well-suited for:

• Enterprise knowledge bases
• Domain-specific AI assistants
• Regulated or high-precision use cases
• Complex business or technical domains

Your ontology becomes the contract between raw text and structured knowledge.

Perseus in the AI stack

Perseus sits between raw data and AI applications:

Unstructured Text → Perseus (Text → Knowledge Graph) → AI systems

Instead of prompting LLMs to “figure it out” every time, Perseus gives them a reliable knowledge layer they can build upon.

Why use the SDK?

The Perseus SDK lets you integrate this capability directly into your applications:

• Programmatic access to the Perseus service
• Easy submission of text and documents
• Full control over ontologies
• Seamless integration with vector stores, graph databases, or custom pipelines

Whether you’re building an internal tool, a customer-facing AI product, or a large-scale knowledge platform, the SDK gives you fine-grained control over how knowledge is created and used.

When it is the right choice

Perseus is a strong fit if you need:

• More than keyword or embedding-only search
• Explainable, structured AI outputs
• Cross-document reasoning
• Long-term, reusable knowledge
• AI systems grounded in your domain model

If your AI needs to understand, not just retrieve, Perseus provides the missing layer.