Where is this thing going?

Where we sit today sits at the low level. AI models sit at a point where full forced analysts can be created to handle singular costs for 100 a month. If I have it run a market survey it'll pull the data and miss the depth, so it takes a second pass, every time. Most of the time isn't the same thing as all of the time, and most of the time isn't good enough when the number is going to an investor. So I treat it as a first-draft tool. It gets me to a starting point faster than I'd get there on my own, and everything after that is still mine.

The ceiling on all of it, in my read, is trust with capital. A lot of the liquidity in this business sits with people who didn't grow up with this technology and don't use it the way somebody younger does. That group will lean on it happily for the analytical work underneath a decision, for pulling data, for a first pass at an underwriting. Handing over the decision itself, and the money behind it, is a different act. I don't see that shifting across the majority of this industry in the next several years. Some will move sooner, and younger capital probably moves first. That's true of anything.

What does move sooner is the layer underneath. Getting AP processed, reporting produced, financials calculated, the first version of a document drafted. I've watched that happen in my own work and I expect a lot more of it.

The part I keep coming back to is that a model doesn't feel anything when money is lost, and it doesn't feel anything when money is made. It's a robot. Everything that actually matters in this business sits on somebody being accountable for a number, and accountability isn't something you can hand off to software.