Giving an LLM your prod database is easy. Taking access away is the hard part
deepsql.aiIt seems they're conflating two different risks? Read-only connection is a solution to Integrity risk, while they straight away start talking about Confidentiality risk instead.
They're also problematic, but different kinds of problematic and I don't see how this is different from managing employees database access.
Then there's no mention of handling Availability risk management. An employee or agent querying the entire order table joined with... and the entire db cluster is choked for two hours.
Data confidentiality and sovereignty matter a lot to enterprises. You may want to redact access to the most sensitive data from both humans and agents, and it gets tricky to enforce such dynamic policies at the database level. DeepSQL solves this by managing all DB users (humans and agents) at the DeepSQL layer.
DB availability is out of scope for DeepSQL at the moment.
Prompt-to-Access Management is really helpful for admins, it is not just a productivity boost but also adds a governance layer that manages both Agents and human identities under one umbrella.
Generated prose is devoid of meaningful semantic content.
The one takeaway I got from this article is that deepsql.ai are the last people I would trust to handle this problem, because they can't even explain it themselves.
Blog post is displayed for a moment and shortly after the post text is replaced with a cryptic error about regex parsing. I guess it has something to do with Safari javascript engine. Anyway, the current state of web development is so pathetic.