The Four Elements of AI Transformation

TL;DR

AI transformation requires systematic change across four elements: culture, organization, process, and technology. Most companies fail at culture because employees aren't invested enough in outcomes to justify changing their behavior. The solution: build elite GTM teams, avoid AI-washing, and be very careful about automating customer experiences.

Why most AI transformations fail

In a recent conversation with Zack Kass, former OpenAI executive, he shared why so many companies struggle with AI adoption despite massive investment. His answer wasn't what I expected: it's not about the technology.

The pattern he described matched what I've seen firsthand: companies buy the tools, hire the consultants, announce the transformation but nothing really changes. Employees keep doing things the old way because they're not invested enough in outcomes to justify the effort of changing.

The four elements framework

Kass identifies four elements that must change for successful AI transformation: culture, organization, process, and technology. Most incumbents struggle because they approach this backwards. They start with technology when culture is the real blocker.

The cultural element is where most companies fail. As Kass puts it:

Most people are not connected well enough to the outcome of the company to justify the effort to changing how they operate.

This is why startups have an advantage as most employees are (more) invested in the outcome. If your employees aren't willing to change their behavior because they're too invested in their "identity," it will not work.

You can look at the framework like this:

• Culture: Are our people invested enough in outcomes to change?
• Organization: Can we move people around and create smaller teams?
• Process: Do we have processes to automate in the first place?
• Technology: What can we commoditize vs. what should we build?

Building elite GTM teams instead of massive orgs

The organizational element requires a fundamental shift in how you think about go-to-market teams. Kass recommends playing smaller:

The smaller the company feels, the more behavioral adaptability it will have.

This means building elite GTM teams rather than larger organizations. More than ever, talent density should supercede headcount. You should pay more per head but reduce surface area because an individual can manage significantly more when they're part of a high-performing team.

I've personally seen this play out in practice with sales teams and Zack touched on this as well. Companies that hire 50 mediocre reps struggle more than those with 10 exceptional ones. The math seems counterintuitive until you account for the cost of bad revenue, training overhead, and organizational drag.

When hiring, he suggests looking for high adaptability in senior sales people. Junior people are willing to learn new tools but lack the experience of knowing how and when to trust a buyer. Senior people who can adapt are gold but watch out for those who say they're willing to change but show signs of behavioral baggage. This should be a red flag.

Finally, he shared that the goal shouldn't be to automate the sales process, but rather to bring efficiency to the process. His recommendation is to think of GTM engineers as employees who should take individual reps from $1M to $5M of managed ARR.

The automation paradox: when AI destroys your brand

Here's where most executives get it wrong. Zack warns:

"Be very careful about automating the customer experience. There is a lot that can be done but there are societal thresholds. Ask yourself: what do we want machines to do?"

Another challenge with AI is that the cost to try things is low, but that means the cost to switch is also low. This creates a retention crisis. Companies are paying reps on deals they shouldn't be comped on, deals where the extra seats will churn, CAC is too high...

As Zack puts it:

Bad revenue is more expensive than no revenue at all.

So the automation paradox is that you can automate more than ever, but if you do it wrong, you destroy your brand and possibly your revenue.

We've all seen companies rush to automate customer touchpoints in the name of efficiency, only to see their NPS scores plummet and churn rates spike. Sure, the technology works perfectly, but the customer relationship is dead.

How to spot AI-washing in your GTM strategy

Everyone in PMM has overused AI and agents, it feels like crypto three years ago. The best PMMs don't talk about AI, they talk about outcomes (and sometimes solutions).

Interestingly, he doesn't feel like we are going towards a product discovery issue but rather a customer experience issue. The market continues to have a ton of information and to surface the best product to prospects. Maybe even more than ever. His theory is that you should continue focusing on making customers happy. Even if your G2 or Trustradius reviews are viewed by as many people directly, a happy customer is still the best source of new business.

In addition to the referral component, with cost of buying and switching coming down, the most important thing you can do is build a great product. He recommends companies focus on retention over acquisition. Some companies are growing slower but with great retention, while others have tremendous speed but high commoditization (and high churn in the future). His bet is that the winners will be the ones who focused on retention.

The uncomfortable questions

There was a lot to unpack from this conversation and it gave me a lot to think about. Though if I needed to take only one insight away it would be the AI transformation framework:

• How do I get my people invested enough in outcomes to change how they work? This starts with hiring and gets reinforced with rituals (customer centricity, comp plans...)
• Which elite teams can we create within the org to move mountains and show the way? Talent density takes time to build but you can create sub-orgs with higher density and start having them permeate the organization.
• Are we automating $5 tasks or $5000 tasks? Looking at this from a value perspective rather than a cost perspective changed the way I think about AI automation.
• Is what you're trying to do with AI so unique that nobody has built a solution to do it? Only build what cannot be commoditized because this becomespart of the IP of the company.

I catch myself daily making the wrong choices intuitively, especially when things are moving so fast. This framework has helped me recenter my thinking and make better decisions.