What is Graphlit?

Graphlit is an API for ingestion, memory and retrieval for AI apps and agents.

Let's break that down...

• API: Graphlit is a cloud-native managed platform, which is accessed via GraphQL API, or native Python, Node.js or .NET SDKs.

• ETL for LLMs: Graphlit automates the ingestion, extraction and enrichment of unstructured data from any source. With built-in metadata parsing, text extraction (including OCR) and vector embeddings, your end-to-end LLM preparation pipeline is managed for you. In addition, Graphlit extracts knowledge graph entities and relationships between the people, organizations, places and topics found in your user's content.

• Knowledge Retrieval: With the recent innovations of Large Language Models (LLMs) and Large Multimodal Models (LMMs), application developers are looking to extract knowledge from multimedia content, ask questions about the content, and search, summarize, and repurpose the extracted knowledge. Graphlit provides the Retrieval Augmented Generation (RAG) pattern of knowledge retrieval - as a usage-based API.

RAG-as-a-Service

As you look to integrate LLMs and LMMs into your application, developers often start with open source projects like LangChain or LlamaIndex. You also need to select a vector database, like Pinecone or Qdrant, and a text embedding model like OpenAI ada-002. Once you've selected the technologies, there is learning curve to integrate them, not to mention how to optimize the integration for your use case.

Graphlit is different. We provide RAG-as-a-service, where you can focus on your vertical AI application, and let Graphlit handle the AI data infrastructure for RAG.

We offer an end-to-end platform, with pre-integrated vector and graph databases, and integrations with all the leading LLMs and LMMs.

Graphlit does let you configure your conversation and prompt strategies, i.e. windowed conversation history or prompt rewriting, but out of the box, no configuration is required - it just works.

Compared to building a chatbot or copilot with open source projects, Graphlit gets you up and running in minutes not weeks.

AI Apps and Agents

Graphlit provides a unique platform for adding long-term memory and retrieval to AI applications and agents, where the benefit to the user is from the knowledge extracted from their complex data, not just in the data itself.

Businesses create and collect many forms of complex data, such as Word documents, PDFs, meeting video recordings, emails, and Slack messages. Not to mention, mobile phone images, CAD drawings, 3D renders and geospatial data. Locked in these files is the collective knowledge of the organization, which previously has been difficult to search, catalog, and integrate.

Embedded knowledge can found outside businesses, as well: Podcasts, YouTube videos, Twitch streams, Reddit forums, and Web content.

Using AI, Graphlit builds a searchable, conversational knowledge graph from your or your user's data. As a platform, Graphlit provides webhooks and a connector model for extensibility, because the value comes from integrating this knowledge with your application.

Developer Focused

Graphlit is for developers building innovative applications using AI, which leverage domain knowledge in any vertical market: i.e. legal, sales, entertainment, healthcare, construction/engineering.

Graphlit lets you focus on your unique application requirements, while providing an easy-to-use API which simplifies complex data workflows, including data ingestion, knowledge extraction, semantic search, alerting and application integrations.