Funding
Funding not disclosed
Founders
Product
Problem
Organizations struggle to effectively navigate and derive insights from disparate data sources, including unstructured files and communication logs. This lack of interconnectedness hinders the ability to understand data lineage, verify information, and gain comprehensive observability into the underlying knowledge represented by large language models (LLMs).
Solution
Hippograph transforms uploaded files, messages, and data into a dynamic knowledge graph, enabling users to navigate information and discover connections. The platform provides enhanced observability into LLM operations by exposing data sources and interconnections through an interactive chat interface. Users can visualize and explore these knowledge graphs via a user-friendly graphical interface, facilitating insight extraction and information veracity verification. The system supports a range of foundation models and allows for custom model fine-tuning with annotation and sourcing.
Target Audience
The primary target audience includes data analysts, AI/ML engineers, and knowledge management professionals who require advanced tools for data exploration, LLM observability, and insight generation from complex datasets.
Features
- Dynamic knowledge graph generation from diverse data inputs (files, messages)
- Interactive chat interface for querying and exploring data interconnections
- Visualization tools for navigating and understanding knowledge graph structures
- Support for multiple foundation LLM models (GPT4, GPT3.5, Gemini Pro/Ultra, Llama 2, Claude 2/3, Mixtral-8x7B)
- Custom model fine-tuning pipeline with annotation and sourcing capabilities
- REST API for programmatic access to the knowledge graph and data enrichment
- Subgraph generation linked to LLM responses for verification and exploration