We open sourced our entire text-to-SQL product

Here's the thing: open-sourcing a natural language-to-SQL engine sounds great, but let's not forget the operational complexities involved. Dataherald claims to empower business users to query relational data without data analyst intervention, but if your data quality isn't solid, this can lead to chaos. You don't want users acting on insights driven from garbage data. Plus, the emphasis on a Slackbot integration is enticing, but the real question is, how well does it handle complex queries? What they're not saying is that relying on natural language processing in production environments requires robust monitoring and error handling — something many tools overlook.

If you're already using tools like OpenAI Codex or Google Cloud's BigQuery ML, consider whether the benefits of Dataherald's components outweigh the integration and maintenance overhead. Each service has its own docker-compose.yml for setup, which is a double-edged sword. On one hand, it simplifies deployment; on the other, it adds complexity in managing multiple services. Who will handle the operational burden when something breaks at 2 AM?

Dataherald's architecture is modular, so you can deploy just what you need. This can be appealing for teams wanting a tailored solution. However, if you're looking for something battle-tested with proven scalability, you might want to hold off. The lack of performance benchmarks is a red flag. Until you see how it performs under load, it’s hard to gauge whether it can handle your production requirements.

In short, while Dataherald is open-source and presents a promising concept, it's still in early general availability. For teams comfortable navigating early-stage software and keen on experimenting, give it a try. But if your focus is on stability and proven results, keep evaluating your options. The catch here is that with natural language querying, garbage in means garbage out, so ensure your data quality is up to par before diving in.