truefoundry/cognita
Cognita
Cognita is a retrieval augmented generation orchestration framework used to build pipelines that connect document stores and language models to provide grounded answers. It functions as a document ingestion pipeline and a vector database integrator, managing the process of loading, parsing, and indexing files into a searchable knowledge base.
The system includes a language model gateway proxy that provides a unified API to interact with multiple different model providers. This routing layer decouples the application from specific vendors, allowing requests to be proxied through a provider-agnostic interface.
The framework covers contextual information retrieval through similarity search and reranking to generate responses with source citations. It supports incremental document indexing to process new or updated files without re-indexing entire datasets and allows for the integration of various vector store implementations.
Features
- RAG Frameworks - Offers a comprehensive orchestration framework for building retrieval-augmented generation pipelines that connect document stores and language models.
- Retrieval-Augmented Generation Frameworks - Provides a comprehensive framework for orchestrating retrieval-augmented generation pipelines, including node sequencing and reranking.
- Contextual Information Retrieval - Fetch document chunks using similarity search, query decomposition, or reranking to provide grounding for responses.
- Contextual Retrieval - Implements algorithms for ranking and fetching relevant context chunks to provide grounded AI answers.
- RAG Pipelines - Provides workflows that augment model outputs by retrieving and integrating relevant external data from document sources.
- LLM Gateways - Functions as a centralized interface for aggregating multiple LLM providers into a single API.
- Model Provider Proxies - Provides a middleware layer for routing and unifying requests across different AI model providers.
- Model Proxy Gateways - Provides a unified middleware interface for interacting with and routing between multiple LLM providers.
- RAG Context Retrieval - Retrieves relevant document segments from a knowledge base to provide factual grounding for language model responses.
- Document Ingestion Pipelines - Extracts, parses, and chunks raw documents into vector embeddings for semantic search and retrieval.
- Vector Database Integrations - Provides connectors for multiple vector store implementations to manage and query high-dimensional document embeddings.
- Model Request Proxies - Acts as a gateway that intercepts and routes API requests to interchangeable language model providers.
- RAG Component Modularity - Implements an architectural pattern that decouples retrieval, encoding, and generation into independent, modular services.
- Grounded Answer Generation - Generates responses supported by traceable citations and source verification from retrieved document chunks.
- Data Ingestion Pipelines - Uses customizable loaders and parsers to normalize diverse documentation formats for vector storage.
- Document Ingestion Pipelines - Ships workflows that parse raw files into structured text chunks and metadata for semantic search.
- Vector Stores - Integrates with various vector store implementations to manage high-dimensional document embeddings.
- Incremental Index Integration - Supports merging new or updated documents into existing indices without requiring a full dataset re-index.
- Incremental Vector Sync - Keeps vector indexes continuously updated by processing only the delta from live data sources.
- RAG Frameworks - Modular framework for building production-ready RAG applications.
- Retrieval Augmented Generation - Modular framework for building production-ready RAG applications.
- RAG Frameworks - Modular, API-driven RAG framework for production-ready environments.
