I find it useful to have goons on payroll that make these kinds of problems go away

I promote people for simplicity

how is this a company?

I don't find this all that compelling. Different technologies can have different effects. And why would future effects be influenced by the accuracy of random people's predictions of other events in the past?

I think mine confabulated criticism of a point of view as espousing that point of view.

don't loving these

We have done a lot of work solving this exact challenge at legaide.ai. I was both sad and relieved when gemini 3 didn't deprecate our startup.

I don't really understand this. You can and should tell the llm the source of the search results.

We often will repeat calls to try again. Or sometimes we make the same call multiple times to get multiple answers and then score or merge them.

Is this used only in cases where you assume the answer from your first call is correct?

I’d love your opinion here!

Right now, we assume first call is correct, and will eagerly take the first match we find while traversing the tree.

One of the worst things that could currently happen is we cache a bad run, and now instead of occasional failures you’re given 100% failures.

A few approaches we’ve considered - maintain a staging tree, and only promote to live if multiple sibling nodes (messages) look similar enough. Decision to promote could be via tempting, regex, fuzzy, semantic, or LLM-judged - add some feedback APIs for a client to score end-to-end runs so that path could develop some reputation

I’d assume RL would be baked in to the request structure. I’m surprised OAI spec doesn’t include it, but I suppose you could hijack a conversation flow to do so

Smart invisible regex makes it fast and prevents it from crashing? What does that mean?