Hacker News Reader

Comment by pxue | Hacker News Reader

Unfortunately, we had a recent discussion internally amongst the execs and have decided to not allow anyone use tools like this.

There's just no way to know for sure how securely data are stored and what possible attack vectors are.

It's a great idea but imho the only path forward is allowing custom deployments against pretrained models.

View on HN

According to docs it only takes in your data schema and outputs the SQL query to run, so the AI model doesn't actually get information about your data, just the schema. Otherwise data is handled according to SOC II principles similar to any other BI tools.

Doesn't this penalize the tool though? Knowing the amount of data in a table is still important to determine the type of query, even more to optimize it

Yeah, in general the more data you're able to use (assuming the context window supports it), the better results tend to be. We arrived at the data schema being a good enough compromise at which the benefits outweigh the risks for several use cases. Besides, some data stores that are generated by third-parties actually have common schemas (think Sendgrid / Hubspot activity data), so you're not risking much but potentially gaining a lot of sales ops productivity.

Db schemas sometimes contain important information themselves, as well as proprietary architecture.

I see it same as slack leaking private channel names.

Reply on news.ycombinator.com