FAQ
What is DeepDive?
DeepDive is a tool for data analysis and visualization.
It takes questions, translates them to SQL, executes them and visualizes the result. See our demo here.
How does DeepDive work?
DeepDive uses natural language models to translate questions to SQL.
DeepDive post-processes and executes translated SQL queries. Resulting data is rendered as visualizations, which users can edit through DeepDive's UIs. Edits to visualizations modify and re-execute underlying SQL queries.
What data does DeepDive support?
DeepDive supports Excel and CSV files up to 50MB.
DeepDive does support other SQL databases (e.g, Snowflake / Postgres), but they are not publicly available as of now.
If you're interested, please drop us a note at dev@bkdevs.io
How does DeepDive store my files?
DeepDive stores files encrypted at rest in S3.
How does DeepDive process my data?
DeepDive loads files from S3 on session load.
A temporary SQL database is then created and used for the session. This database is deleted on disconnect (i.e, tab closed, user navigated away). Your files are not persisted in any intermediate form otherwise.
What models does DeepDive use?
DeepDive is currently using OpenAI GPT-3.5
If you're interested in using GPT-4 or alternative LLMs, let us know: dev@bkdevs.io
What data does DeepDive send to OpenAI?
DeepDive sends table metadata to OpenAI to generate SQL queries. This comprises of table names and column names.
We do not send your data to OpenAI.
Who built DeepDive?