Datasaur

Datasaur is a data labeling platform designed to manage the entire data labeling workflow for natural language processing (NLP) and large language model (LLM) projects. It uses both AI and human annotators to label data and allows developers to assign numerous annotators for specific topics, reducing subjective biases. The company offers its platform through a subscription model that includes a free tier.

The company also offers a tool called Dinamic that allows developers to train and build custom NLP models by using data annotated on its platform. Dinamic autonomously refines its learning as additional data gets labeled, enhancing the model's accuracy. Moreover, it introduced a new feature (in June 2023) that allowed customers to label data and train their own customized ChatGPT model. It also enables human annotators to analyze the quality of the LLM outputs and determine whether the responses meet specific quality requirements, thereby expediting the model training and building process by incorporating human feedback (reinforcement learning).

Datasaur also offers a platform for developing and training custom LLMs called LLM Lab. This integrated interface enables developers to create custom GenAI applications, providing features like internal data ingestion, data preparation, augmented generation, model selection, and optimization of similarity search. The platform is available for both cloud and on-premise deployments.

Moreover, in March 2022, the company acquired Konvergen AI, an optical character reader (OCR) technology startup. Before the acquisition, the two companies had collaborated on several projects. Following the acquisition, Datasaur planned to integrate Konvergen AI's specialized technological capabilities in handwriting recognition, government ID field extraction, and intelligent document processing.

Key customers and partnerships

Datasuar partnered with Consensus, an AI search engine for research, to assist the latter to annotate scientific papers.

Funding and financials

In August 2023, Datasaur raised USD 4 million in seed funding, led by Initialized Capital, with participation from HNVR, Gold House Ventures, and TenOneTen. It aimed to use the fresh funds to advance its NLP (natural language processing) data-labeling and model-building capabilities.