Datatune

While building Agents for a customer who had large amounts of data, we saw that their Agent struggled with certain data transformation tasks, which would have performed better if LLMs had access to the full user data as well. We built Datatune as a first step, to solve this issue. With Datatune, users can: - Work with their data with a row-level intelligence of LLMs, Agents, or Embeddings. - Use a wide variety of Data Sources such as Databases, DataFrames, etc. - Perform tasks such as map, filter, semantic deduplication, etc with just a simple API. - Let Datatune Agents decide the data transformation architecture required to perform the user prompt.