Pinsight is building open source large language and vision language models (LLMs and VLMs) that are able to conduct analytics and extract insights from both structured and unstructured/scanned documents.
TOLMA (Training and Optimizing Language Models for Analytics) is a state of the art LLM that can analyze tables and produce reports, charts and maps.
DOLMA (Document Optimized Language Model for Automation) is a multi-modal LLM that can produce structured JSON file from scanned document images.
With commercial solutions like ChatGPT and BARD leading the market, we oftentimes are asked why should someone prefer our models over commercial giants.
Our models are open source and hosted publicly. You can use them for free and customize them to your needs. This level of customization is not provided by commercial models.
ChatGPT and BARD are great generic assistants. Pinsight is building narrow, specialized expert assistants that excel in data analytics and reading scanned documents.
Our models can be hosted behind your firewall and meant to be used in privacy preserving setting ensuring complience level required by GDPR, SOC2 and HIPAA complient companies.