Build vs. Buy with Andy Quick: How Asking the Wrong Questions Leads to the Wrong Answers

Every utility exploring AI applications eventually hits the same fork in the road: build a custom solution tailored to your specific systems and workflows, or find a vendor with something turnkey and get moving faster. It sounds like a technology decision. In practice, it's much more complicated than that.

Here at Noteworthy AI, we recently welcomed Andy Quick as a Senior Industry Advisor, bringing nearly three decades of utility experience to the team, including his tenure as Entergy's first-ever Chief AI Officer. In that role, he built the internal framework for how one of the country's largest investor-owned utilities could create measurable value from AI at scale. Now he's bringing that perspective to the utilities we work with, helping them cut through the noise and make smarter technology investment decisions.

In a recent episode of the Factor This podcast, host Jeremiah Karpowicz sat down with Andy to get his take on the questions utilities are asking, and the ones they should be asking before they even start the build vs. buy discussion. Keep reading for the highlights of this conversation, or check out the full podcast episode below.

‍

Q: Looking back on nearly three decades at Entergy, was there a single most disruptive shift you saw in the industry, or was it more gradual?

A: Before AI came along, the transition to clean energy seemed like the big disruption. Then AI walked into the room and became number one from a disruption perspective, especially with the AI data centers coming in. The ability for utilities to respond to that demand from hyperscalers, from both a transmission and generation perspective, is something we've never seen before. Despite that growth, I think the adoption and integration of AI tools into utilities at scale is going to be gradual. It's going to take quite a number of years.

Q: As Chief AI Officer, you formed a new AI department focused on value creation and governance. What was your approach?

A: What we tried to do was answer three critical questions. Number one: how do we create material value with AI? Number two: how do we enable the organization to take advantage of these tools to improve productivity? And number three: how do we make sure we're not running with scissors, meaning how do we manage the potential risks?

There were really three pillars. One is top-down: what are the big opportunities to use AI to make a significant difference to high-level company objectives? Two is bottoms-up: how do we create organic innovation from everybody in the organization, because people come up with really cool ways to use AI. And three is making sure we do it in a way that mitigates the risks associated with these models.

Q: Does it make sense for utilities to start an independent AI department, or should AI live within individual departments?

A: The answer is anything can work. It could be in the IT function, an independent AI function, scattered across departments, or a center of excellence. On paper you can come up with all kinds of frameworks, but that's not what's most important. What's most important is where in the organization you want to cluster AI capabilities so it has the greatest ability to influence change, because creating material value equals change.

Some of these processes at utilities are very old and institutionalized. Going from how we've done something for 20 years to doing it completely differently with AI is the hard part. So the question should be who in the organization, at a high enough level, can influence that level of change, both to win people's hearts and minds initially and to sustain it over time. You want to find your champion, because the change is the challenge, not the technology.

Q: That sounds like it ties directly into the build vs. buy question utilities face. What's underestimated on the utility side when making that decision?

A: I always like to back up the bus a little bit. Before you even enter into buy versus build, have you found a problem worth solving? Do you understand the level of disruption and change that's going to go into that? And do you have the support from a change management perspective to make that change happen? Because if you don't, it doesn't matter whether you buy or build. It's kind of a kill switch in my mind.

If you get to the point where yes, this is a problem worth solving, we understand the change required, and we've got good sponsorship, then you can filter it through a few things. Number one is speed to value. I always put speed at the top. Number two is the economics. Has the market already solved this problem? If there's a company with a solution, they've spent millions of dollars and probably years building that platform. You will take at least that long and at least that much money to reach feature parity, and by the time you get there, they've moved on and their solution is even better. So it doesn't make sense to build your own when the market has already answered for it.

Third is integration. That can really bite people in the back. Here's this nice tool that does all these things and sounds great, but utilities have many legacy systems. How is it going to really fit in? I always advise talking about integration with a provider up front. Now, if the problem hasn't been solved by the market, you're in novel territory, and I'd always recommend partnering with somebody if you go down that path.

That's actually one of the reasons I teamed up with the Noteworthy AI team. It's visual inspection. It's already pre-built. You stick the camera on a truck and it integrates with everything. It lights up your GIS and you can start doing inspection right away. There's nothing really novel about it, so building your own in that case doesn't make sense. Customer service is another good example. There's a lot of money going into customer-facing interactions, and I wouldn't recommend a utility build their own solution there either, because that problem is already being solved by many players in the market.

Q: Where do the champions who drive this kind of change actually come from inside an organization?

A: Finding those champions is more art than science. The premise is that to realize material value from a discretionary investment in AI, you need a senior leader who is very passionate about solving a particular problem. Somebody high enough in the organization, banging their fist on the table, saying we've got to solve this. That is the critical success factor.

I get asked all the time why AI projects fail and how to find a winning use case. It's two sides of the same coin. It's finding that leader who really has the passion to solve the problem. That's your gateway to a successful AI investment. Otherwise you end up with a bunch of small use cases that become experiments and fail, because the amount of value is directly proportional to the amount of change. If you're just making one step in a process a little easier, that's not going to have significant value.

This happens at every level, not just at the top. I know engineers who've had brilliant ideas but were in organizations where they had no influence to act on them. That's where an AI function within an organization can help. I always recommend having that bottoms-up mechanism. You're giving employees tools, but you're also creating a community, a structured space where people with great ideas can promote them to the executive team. Otherwise those ideas just sit in the queue and never get attention.

Q: How should utilities think about the full economics of build vs. buy, including opportunity cost?

A: I love that question because I always go to opportunity cost first. The pathway to success is through disciplined prioritization. I like the model of desirability, viability, and feasibility, in that order. Desirability is whether somebody senior in the organization wants to solve this problem badly enough to drive the change. If you don't have that, it's a kill switch.

Viability is the economics: do the benefits outweigh the costs and risks? And last is feasibility, the technical stuff, the data and tools and models. Ironically, that's where a lot of people want to start. It's the tail wagging the dog. I almost never see technology being the wall. People feel like they need to get all their data straight before they can do anything, and I'd encourage not thinking that way. Assume you can figure it out case by case, but don't try to boil the ocean.

Take a moment to chart out what the journey looks like if you're buying versus building, including upfront cost, ongoing cost, and any integrations required elsewhere in the organization. If the market has already solved the problem, the buy path is very likely going to prove more economical, because that vendor has already made the investment.

Q: Culture and change are hard to quantify. How does that fit into the opportunity cost conversation?

A: It's number one, honestly. I've spent my whole career integrating technology with business process and change, and it's the same lesson over and over. Value comes from change. AI is just another tool in the tool belt, but when you boil it down, you're transforming process with technology to harvest material value, and that requires change. If you sweep that under the rug, it will come get you. If you confront it up front, before you take the first step on a project, you're maximizing your chance of success.

People who think they need all their data straight before they can do anything are setting themselves up for an eight-year wait. Bad data is created by humans, and as long as there are humans hands on keyboards, bad data is never fully going away. That's a legitimate concern, but ask yourself what you're going to say in a couple of years when the answer is that you're still cleaning up data. There are plenty of use cases where you can find good enough data, strong executive support, and a clear path to value. Think of desirability, viability, and feasibility as a Venn diagram. You want to hit all three.

Q: What development or trend in the energy sector are you most excited to see take shape this year?

The one thing I'm really watching is getting AI into the field, especially in distribution. Like Noteworthy AI, they've got a solution that requires fairly little integration. You can start using AI for inspections right away. Beyond that, there's a great opportunity to integrate AI into the day-to-day work of distribution employees to make their lives easier and more efficient. There's a lot of administrative tasks in utilities, and AI can help workers spend more time on hands-on work. Change happens extremely gradually at utilities, so I don't think you'll see a switch flip in a year, but the opportunity is significant.

The other area is the regulatory space and efficiency in rate-making, which I think is a big opportunity for both utilities and public service commissions. From an employee perspective, getting AI into the field is also about lowering the amount of non-field work, like typing into systems and maintaining records, so people can spend more time doing the line work they were actually trained to do.

Q: Last question. Tell us something powerful, whether it's specific to energy and utilities or more broadly.

Think big. Think really, really big, and don't be afraid of the unknowns. If you really want to make an impact within your utility, assume that AI can do anything you want it to do, but think very big about the kind of problems you want to solve. Shoot for the moon. Don't think small. Think big.

To learn more about how Noteworthy AI can help your utility move from evaluation to deployment, reach out to our team today.