Karl Yeh

I don't know if anyone knows Trimble, Invista, Innate, Vista, those type of systems.

Then in, in accounting, you have different types of systems too, like iFirm or Caseware.

Like,

So if each industry has these type of systems related to it, how are you going to connect your AI into any of these systems?

Now, you could MCP it, but that's very difficult if you're trying to build MCP for all the tech stuff, every single software.

But the thing is, if you create a data lake,

you have to be very careful that it doesn't become this catch-all and it's now hard to navigate into it.

You have to be very, very, very structured.

So I think for us, one of the first things that we would recommend is instead of creating a major data lake and doing everything,

pick the departments that you want to start with, and then create, like, a very thin, like, data warehouse to test that first.

And most companies, you have, like...

a Microsoft Fabric that you could build an MCP on top of.

You could BigQuery this.

You could look at Snowflake or Databricks to start.

That's where you do it.

I think which would, unless the other thing that you could do is what a lot of companies do because none of these stacks talk to each other.

The number one thing is they export it to Excel, work it, and then import it

to a different system via Excel.

So all the work is still done on Excel.

It's just imported export to different systems to ensure they talk to each other because the APIs don't connect.