Josh Daniell

It's a question for AI generally.

Will it stay inside the box that we create for it as humans?

But yeah, absolutely.

If there was a sense that this can't be trusted, then people are not going to use these services.

Yeah, it's a really interesting point.

I think if you hear what people from the AI labs say on this front, they will say that prompt engineering has an error and that error will fade because the AI is getting good enough to try and understand the intent or what is actually good for the human instead of relying too heavily on the specific words that are used in a lazy, quickly written prompt.

So you hear Anthropic or OpenAI people talk about this and they'll say, yeah, prompt engineering was a thing, has been a thing, but it will fade away and how important that is for these kind of queries.

Another thing is how these models are tuned for personal finances.

And so we were talking earlier about OpenAI's acquisition of Hero and Perpixity's product integration with Plaid.

Those are examples of some of the key AI services globally making a leap into personal financial management.

And one of the things that Perplexity noted with this announcement is that over 75% of their paying users ask a personal finance question every month.

So people are turning to AI for help with their personal finances.

So these AI services are

are trying to meet their users exactly where they are.

And what I'm trying to say is that there is intent to make the AI work really well for the scenario.

It's not an ancillary kind of part of why people are using these services.

It is one of the core parts of why people are using them.

So what we're seeing is this real intent to lean in.

Exactly.

Well, everything, because I love to know what is out there.