I talked to Lucas about what companies need to do to not just survive, but succeed in the AI era. This conversation has been edited for length, clarity and continuity.

How do you recommend a company get all of its data sprawl, apps used by various employees and departments, and legions of APIs under control?

Lucas: The first thing is acknowledging you have a problem. The first thing is recognizing the digital fragmentation—a term I use frequently—exists and it is a real problem.

The second thing is then assessing within your organization: What are the systems, the applications, the data, the business elements that I need to run and operate my business? What are the core things that I must absolutely have on a day-to-day basis?

The third is mapping the processes within your organization, what I characterize as the hidden processes today. So think about this for a minute. How many CEOs could say: ‘I know exactly how our income statement is assembled at this company… all the systems required to pull that data together … the spreadsheets that are sitting out there with the magic translation that my accounting team does.’

Knowledge processes are very different than business processes. A business process is a manufacturing process, and I know every step that goes into the assembly. The knowledge process: Do I understand what elements go into the assembly of the income statement as a product? Most organizations don’t have a good model for what their knowledge processes are, so you’ve got to inventory that.

Once you do that, you have the ability to weigh: Here’s my knowledge processes. Here’s my business processes. What systems and applications do I really need to achieve this?

Ultimately, where we’re going with this is probably 75% to 90% of the knowledge processes that we rely on today, that human beings work on, will go to AI.

How does the transformation to AI impact contracts you already have for SaaS and with other tech vendors?

You can’t rely on your suite anymore—rigid architectures, closed systems. By the way, these vendors know that they’re in deep trouble. Your competitors, the ones that aren’t weighed down with these rigid architectures, are building flexible, agent-driven, highly composable systems.

You have to learn how to extract value from your existing stack, not invest in your existing stack. Companies are not going to win by replacing their core systems. Those are very, very expensive. I go back to a simple example: Hundreds of thousands of companies all over the world rely on antiquated billing systems that are 20, 30 years old. If I want to go in and have my billing system be more intelligent so it’s not sending collection notices to my most important customers, the average company today has to spend tens or hundreds of millions of dollars upgrading their infrastructure just to do that. That doesn’t make any sense at all.

Keeping your existing technology, but making it more composable and more flexible with AI, that’s where this stuff is going.

The last thing I would say is silos are a massive liability. If you’ve worked really hard to create this stack that is siloed in nature, when you hear things like your competitors are moving more quickly, they’re more nimble, it’s because they’ve invested in an integration automation orchestration platform.

What advice would you give to a CIO who is working toward bringing in AI agents and wants to make sure they’re going about it the right way?

Prioritize integration, automation and orchestration. Companies that do, that can build modern composable, AI-driven workflows, will win. Build your digital nervous system early. Invest in a platform layer that lets systems, data and agents communicate. Without that layer, your AI is blind and isolated; you’ll build terrific AI that is totally unable to orchestrate meaningful workflows.

I don’t think we’re in a ‘rip and replace’ world. I think we’re in a ‘wrap’ world, where we can wrap our silos and our systems in intelligence and connectivity. Those systems that you already own, wrapping them in intelligence and connectivity is extraordinarily transformative.

Lastly, create out of the gate your AI or agentic governance strategy before the chaos arrives. I met with a hospital network, and they said, ‘Steve, we all have the technology now to build an amazing AI that could help a doctor look at test results and go, ‘Holy cow, your creatinine level was super high,’ and then it could figure out [potentially related conditions the patient was] here for once upon a time. What we can’t bring together is our digital past. The data for that sits in a hundred different systems, and how do we also operate and access all of those systems in a highly regulated environment where HIPAA still matters? How do we protect your privacy at the same time?’ As compelling as that future is that has arrived, we have to rationalize it with our digital past.