EP 34: Rahul Kanodia CEO of Datamatics: What Real AI Transformation Looks Like Inside an Enterprise

I feel like we talk frequently on the show about the concept of the incrementality with innovation.

And I really respect the ability to be able to, you know, build like start small, build something incrementally, see where it goes.

You don't necessarily have to bias towards having a top down strategy.

You can bias towards incrementality.

So, you know, from your perspective, I'd be curious if there are any other areas in which you feel that there are common misconceptions about AI amongst the enterprise community today.

I think there are some misconceptions.

So one is that they think generally customers feel that AI is perfect.

And it's about 200%.

100% accurately and perfectly.

And it's powerful.

It's still very nice.

It's still evolving.

And it makes errors.

So it can't be perfect.

What I tell our customers is that, look, if you've got a 20, 30, 40% productivity gain, that's great.

Take it home.

Don't expect 100%.

So that's one.

The other issue that they... struggle with is they try to challenge the intelligence of AI.

So they'll give the most difficult project and say, let's see if we can solve it.

But hello, even humans struggle with it.

So my suggestion is take high volume, low complexity, simple stuff, automate that.

When it's high volume, that's when you have maximum financial savings.

Yeah.

And not when you do one-off exotic little problem solving, right? So very often they get into a different mode of trying to solve complex problems and that's the wrong strategy.