AI
by Nasdanika
AI provides artificial intelligence capabilities operating on top of other Nasdanika resources, specifically resource sets which are collections of interconnected models. It abstracts AI components from low-level implementation details.
Last updated: N/A
What is AI?
AI provides tools for artificial intelligence, including processors to describe model elements, generate embeddings, perform semantic search, and build chatbots. It leverages resource sets and models to abstract AI components from low-level implementation details, allowing for capability-based reasoning and graph-aware semantic analysis.
How to use AI?
The AI tool can be used through its CLI, which offers functionalities such as generating embeddings, creating and updating vector stores, running a semantic search HTTP server, and building chat applications. It supports integration with OpenAI and Ollama for embeddings and chat completions. It can be used to chat with a site or a model, and potentially agents based on the Invocation flow can be added later.
Key features of AI
Narrator processors for model element description
Embeddings generation (OpenAI, Ollama)
Vector store integration (hnswlib)
Semantic search capabilities
Chat completion functionalities
CLI for vector store management and search
Chat Vuejs component
Use cases of AI
Explaining relationships within a model (e.g., family relationships)
Adding descriptions to model elements for enhanced understanding
Semantic search and RAG with graph distance awareness
Building chatbots that understand context and relationships
Chatting with a website or a model
Generating embeddings for search documents
Creating and updating vector stores
FAQ from AI
What is a resource set?
What is a resource set?
A resource set is a collection of interconnected models that abstract AI components from low-level implementation details.
What is a Narrator processor?
What is a Narrator processor?
Narrator processors describe model elements and their relationships in multiple ways, enabling explanations based on the model's structure and capabilities.
What vector stores are supported?
What vector stores are supported?
The current implementation utilizes hnswlib for vector storage and retrieval.
How are embeddings generated?
How are embeddings generated?
Embeddings can be generated using OpenAI or Ollama.
What is the purpose of the CLI?
What is the purpose of the CLI?
The CLI provides tools for managing vector stores, generating embeddings, running semantic searches, and building chat applications.