You have 1 article left to read this month before you need to register a free LeadDev.com account.
Estimated reading time: 8 minutes
As AI continues to evolve, CTOs see their jobs becoming more strategic than ever.
As generative AI rapidly reshapes engineering organizations, CTOs have had to guide the way for their teams. Ravi Pal, global CTO at advertising firm Ogilvy One, has been spending the majority of his recent weekends head-down in Claude code, trying to understand how far his team can take the tool.
“I can’t preach what I don’t practice,” he said. “So if I’m preaching to my teams or my leaders that this is real, or I’m going to my clients and telling them ‘this is how you do this,’ I must have tested it and validated it.”
But the impact on their role has been personal in many ways, too. CTOs are incorporating AI into their own workflows, using it for everything from rapidly reviewing codebases to strategic planning and getting feedback. They’re also reimagining the mandate of their roles and what CTOs must act on and deliver in the AI era, both now and in the future.
“AI really makes chief technology officers and CIOs become the chief problem solvers,” said Jason Birmingham, CTO of global financial technology company Broadridge. “To be good at using AI, you have to really be structured in your thinking, be able to really clarify your thoughts and questions, and structure them in a way to maximize the AI. I think technical people are quite well-trained on how to do that.”
AI for code reviews and technical research
While Pranava Adduri, CTO and cofounder at security and governance platform Bedrock Data, said he rarely uses AI coding tools for coding, he’s leveraged them in several other ways. He uses Cursor to generate test cases, OpenAI’s Codex for code reviews, and overall, said these models help him “to rapidly understand complex code bases, evaluate the impact of code changes, and iterate quickly.”
Shannon Weyrick, CTO at NetBox, a documentation and automation platform, is seeing similar gains by incorporating AI into his processes for reviewing code bases. He prompts ChatGPT to provide a summary and says it’s “pretty good” at searching for and compiling the high-level information he needs, like flagging security issues or identifying what a particular piece of code does.
For some, AI acts as a technical research tool, speeding up due diligence. Birmingham, for example, discovered that he can prompt a large language model (LLM) to find user feedback on different tools he’s contemplating adopting to help him vet the options.
“It can find forums where people have been talking about the tools, and it brings back things that are more than just standard feature comparisons,” he said. “It has caused us to go back and ask the vendors some different questions before we made decisions to buy.”
Your inbox, upgraded.
Receive weekly engineering insights to level up your leadership approach.
The always-on brainstorming partner
Almost universally, these CTOs are using AI for critical feedback on strategic decisions and communications.
Pal said he’ll prompt an AI model to be a “judge” and tell him what’s wrong with his idea. Sometimes it’s regarding how to strategize a presentation, and other times it’s about the best way to approach an architecture he wants to build. “It will give me such a reasonable answer,” he said of how AI answers simple technical questions.
Birmingham regularly uses AI models as a “devil’s advocate” when thinking through customer conversations and various strategy decisions, asking it to give him counterarguments or point out flaws in his logic. Overall, he said it’s very insightful in these instances and often points out things he hasn’t thought of.
For example, the company was recently trying to land a deal that would unseat a competitor. When thinking about their pitch and strategy, he went to the AI – giving it context on the competitor and the customer’s perspective – and asked it to present arguments against adopting Broadridge’s software solution back to him. In addition to more obvious considerations like the costs of switching providers, it also found links between the software competitor and the potential customer on LinkedIn and suggested there might be relationships at play that he should be aware of. Additionally, it reviewed feedback from the competitor’s users and, based on that, suggested an angle that allowed him to change the pitch to “get into a crack in the argument.”
“It was very useful in kind of turning the mirror around and saying: here are some of the things that might be happening between the customer and the competitor in terms of their dialogues. Obviously, it’s not in those conversations, but it was able to piece together some things I hadn’t really thought of regarding that, and then it gave me some questions to be ready for,” he said.
Weyrick believes using LLMs in this way isn’t a replacement for having conversations with peers, but rather another avenue that offers unique benefits.
“It feels more like a private conversation. So you’re not worried about the social aspect of maybe divulging too much information. You can give all the details that you feel are necessary. You don’t have to worry about boring them, or keeping them on the line too long,” he said. “It really just feels like you have a private consultant sitting at your desk that’s ready to answer all your questions, or at least point you in the right direction.”
More like this
AI accelerates solving CTOs’ people problems
At its core, the CTO’s role is about managing technical teams. AI is introducing new people management-related challenges for CTOs to navigate, but it’s also helping them solve team challenges, too.
Weyrick, for example, shapes his organization according to the Team Topologies framework, an organizational approach for designing technology teams that was first laid out in a 2019 book by the same name. To further aid him when thinking through specific team structures and management questions within the framework, he also now uses ChatGPT.
“It essentially almost gives me access to the author, right? Because ChatGPT knows about this book and about the sort of industry that’s formed around it, and so I can ask questions and start giving it context that’s specific to our business and get recommendations,” he said.
This process helps Weyrick determine how to assign engineers to different projects. He inputs details about the business goal, what they’re building, and the domain areas of expertise they have, and ChatGPT then provides input on the best shape for the organization and what types of communications should exist between teams to ensure they can deliver. For example, when building the company’s new assurance product this year, it helped him decide who he needs on the operational side versus the software side, as well as divvy up ownership of the cloud work versus on-prem work.
“There’s never enough resources to go around in the startup. It’s common, right?” he said. “And so we have to be able to optimize what we have.”
Facing AI skeptics
One challenge Birmingham has faced is how to position AI adoption among an industry narrative that swings between “AI’s going to make you a superhuman” versus “AI’s going to replace you.” Because of this tension and the potential of what AI can unlock, he believes the people strategy element of the CTO job is now “radically different.”
“I think in some ways it requires more empathy. I think it’s going to require a lot more forward thinking on what I do with my human capital,” he said.
This often means bridging the gap between AI believers and skeptics, which Pal said is the most important duty for CTOs today. To do so requires recentering pro-AI engineers when they pitch AI solutions; he always asks if the outcome of the idea is rooted in ROI or just something that sounds interesting to try. This helps the team align and avoid “shiny object syndrome.”
“I tell skeptics the reality and where we need to move with the proof points, so that we don’t get stuck in the past,” he said, describing how he approaches the other camp with leveled expectations, making clear he’s basing decisions on results and not hype. “But at the same time, I make sure that people do apply the logical lens of, ‘why do we want to do this?’”
Strategy and superintelligence
As AI and its use cases continue to evolve, CTOs believe their roles will as well.
“Longer term, the CTO’s role is less about directing day-to-day implementation and more about strategy,” said Pranava. “The core job becomes deciding which problems are best solved by our skilled engineers, which can be accelerated with AI tools, and which parts of the stack we should just buy. It’s about designing the most efficient system of people and technology.”
If the idea of a superintelligent AI or “artificial general intelligence” (AGI) becomes reality, it could open up new possibilities. In this case, CTOs’ strategic mandate would also have to expand to aligning the superintelligent AI toward business goals, Pranava said, referring to the field of thought that AI models need to be encoded with a human’s (or in this case, a business’s) values. This will become paramount with AGIs, as some fear it could act on its own will. Additionally, he believes skills around systems-level thinking and a mindset of rapid learning would be among the most important qualities for a CTO to possess in a superintelligent AI world.
Birmingham said he’d be lying if he didn’t feel nervous about the possibility, but overall, he’s excited about the potential for transformation it’d bring. Pal, on the other hand, currently suspends thoughts on superintelligent AI to avoid ruminating on worst-case scenarios, he said. But at the same time, he believes a catalyzing shift is in the near future.
“I do believe that in many ways, in two to three years, we’re looking at very differently-wired organizations,” he said. “And the roles each one of us will play will have significantly changed.”
