I'll sometimes use agent swarms for triaging quickly through a number of possible issues. For example, a single request for a code review to uncover all possible issues, followed by an agent swarm to dig deep into each issue to confirm or mark false positive.
I have to disagree with the fundamental premise here. LLMs do indeed have an internal "concept" mapping - the embedding layer. There's some fun demos on youtube[1] where you can manipulate embedding vectors (such as king - man + woman = queen), and these manipulations are done entirely in the embedding/concept space.
You can use something like flyway on top of your existing git/cicd stack. Write the query as a migration, have it reviewed using your git code review process, and merge to run the migration.
It doesn't even need to be a third party. Documents you make for yourself but never send to a third party could still be seized in a criminal context or subject to discovery in a civil one.
Attorney-client privilege is a special carve out because the courts have recognized that clients need to be open and honest with their attorneys to get proper counsel and representation.
This ruling is the court declining to extend that special carve out to non-lawyer AI tooling, and keeping the status quo of contemporaneous documents made by someone discoverable, whether or not shared with a third party. The judge draws from the TOS as an admonishment, effectively saying (my words, not the judge) the TOS should have put you on notice that you have no expectation this data is confidential.
Socials: - github.com/tyzoid
---