news.volyx.in

Show HN: We open sourced our entire text-to-SQL product (github.com)

464 points by aazo11 · 813 days ago · 144 comments on HN

Article summary

The company behind Dataherald, a natural language-to-SQL engine, has open-sourced its entire product. Dataherald allows users to set up an API from their database that can answer questions in plain English. The repository contains four components that can be used together to set up an end-to-end Dataherald deployment. The company is open to contributions and has benchmarked performance with self-hosted LLMs.

Main themes

  • Open-sourcing
  • Natural language-to-SQL
  • LLM integration
  • Enterprise adoption
  • Data governance
  • Self-hosting

What commenters say

  • The decision to open-source the entire product may be a strategic move to gain traction and facilitate enterprise adoption, despite potential risks to the company's business model.
  • Open-sourcing the product can help address data governance concerns by allowing companies to self-host and maintain control over their data.
  • The use of LLMs in Dataherald may raise concerns about syntax and semantics, but some argue that these issues can be addressed through fine-tuning and symbolic lint phases.
  • The company's move to open-source its product may be a sign of financial struggles and a desire to leave a legacy, rather than a purely strategic decision.
  • The open-sourcing of Dataherald may not necessarily lead to widespread adoption, as the product's complexity and requirements may limit its appeal to certain use cases and companies.
  • Some argue that open-sourcing a product can actually increase its value and attractiveness to potential users, by providing transparency and flexibility.
  • The choice of license, in this case Apache 2.0, is important for determining the terms of use and contribution to the open-sourced product.
  • The performance and latency of Dataherald, currently around 20-30 seconds, may be a limiting factor for some use cases, but the product is already being used in production by some organizations.