Unlocking the AI Paradox 💻

I give on why there's this fundamental paradox happening in AI, where the technology works incredibly well, but the vast majority of companies are failing to capture that value.

I'm going to show you what's happening and what you can do to do things differently.

Foreign you've probably all seen this MIT study.

95 of Gen AI pilots are failing.

Many people draw the conclusion from this that AI is bad and it's not worth pursuing yet.

the technology is not mature enough.

Let's just wait.

But that's a pretty scary proposition because that 5% that's working is working really well.

And these companies are seeing massive returns, millions of dollars, tens of millions in some cases.

And so the question isn't really whether or not you should adopt AI, but rather how to be in that 5% instead of that 95%.

MIT says these are the five things that are going wrong.

that we're not building systems that are learning or adapting.

There's a poor workflow fit that you should always build with partners.

I actually disagree with most of these.

I do think there's often a misplaced focus on marketing and sales opportunities when back office is often where the ROI is happening in AI.

And I do think that there are often problems with the UX that prevent users from even wanting to engage with the tools that get built.

But at a much deeper level, I think the real issues are the following.

One AI is new.

Just like in the mobile wave back in 2009, 80% of these projects are going to fail.

That's normal, that's fine.

You need to design a system around that.

Second, skill issues are a real thing.

Vibe coding is not easy.

It takes 20 plus projects to get good at it.

There's a lot of stuff to learn.

but you don't want to put the wrong builder on an important task at the beginning.

You have to find the right person basically.

Third, we see a lot of design by committee.

We see a lot of CEOs designing features and there are a ton of great ways to get hands on with this stuff, but that's not the way The right way to do this is to set strategic priorities and connect the right people and finally clear the red tape.

We're still seeing a lot of organizations with IT and security policies that are way behind the times.

And there are great mitigations for these that keep your data secure, but allow people to build the tools that they need rapidly.

So how do we mitigate?

First, we respect and acknowledge the fact that things that used to take six months can now be built by one person in a few weeks.

that's a fundamentally different way to think about things.

Second, we find that operator engineer, that person who has the rare mix of business acumen and technical ability curiosity to get these projects done.

You really only need one person to do that, and they can have an incredibly outsized impact if you enable them properly.

Third, we try to run our AI bets like a VC portfolio.

We design the system so that it's okay when most of them fail.

Make small bets fail fast.

And then finally, we make sure that execs are setting priorities, strategic priorities, and connecting the operator engineers with the end users and clearing the red tape by partnering with IT to create development infrastructure that empowers people.

These aren't complicated changes, but they do require require a, different mentality and that's what this is all about.