Hey HN, Hope everyone is doing great! I have built a fairly early stage prototype of something called as a Data Infrastructure Assistant - https://render-dda.onrender.com/.

Try the following prompts - ‘What is the Calculation behind metric TotalRevenue’ - ‘Calculation behind metric TotalRevenue’ - ‘In which table is the column TotalAmount’ - ‘Is there a table that contains details of an order from a customer’

Please see the problem it aims to solve and the solution,

Problem : Quick and easy access to metadata (data dictionary, metric definitions) has been observed to be a challenge for data teams.

Solution : With Data Infrastructure Assistant, we aim to solve the above problem (think Google/Perplexity) for your data infrastructure. Vision is, to provide quick and easy accessible information to the data teams.

Under the Hood: Leveraging LLMs to access metric specific information. Description for the metric definitions is pulled off of the code written to generate the metrics. Also, this way the engineers don’t have to spend as much time on documentation.

Is the problem real? Happy to get any feedback on this!

Thanks and cheers!