CEOs want their CIOs to move faster on AI without putting the business at risk.
AI-powered competitors could undercut established businesses, putting pressure on CIOs to build their teams' skills and move useful applications into production. Yet faster adoption also raises questions about what AI should be allowed to do on its own, especially when customers, money or sensitive data are involved.
The pressure comes as many companies still struggle to deliver substantial results. Only 13% of CIOs said their enterprise has realized significant or transformational value from AI, according to Gartner survey findings released this week at its IT Symposium/Xpo in Orlando. For CIOs, that adds to the challenge of moving quickly while making sure AI projects are worth pursuing.
To find out where business leaders want speed and where they want restraint from IT, TechTarget asked CEOs, founders and other business executives directly. Their answers called for faster training, experimentation and internal automation, with tight controls over customer-facing actions and caution about becoming dependent on AI vendors.
Here's where they would accelerate AI adoption and where they would hit the brakes.
Where do you want your CIO to move quickly, and where cautiously?
Understanding the security piece must be of paramount importance all the time, but beyond that, CIOs need to understand -- and this is very vertical-specific -- but in our vertical, we view [AI] as a threat to our business. So, if there's a firm out there that we don't even know about offering an AI-native solution, coming to our clients and saying, 'We can do what they do for half the price,' then we need to move faster than we're comfortable with.
CIOs must find a middle ground between safety and moving faster. It's a muscle CIOs need to develop that they've never been asked to use before.
-- Mike Sullivan, co-founder and chief growth officer, OneDigital
You must move faster at building your AI capabilities within your IT team. In reality, there are skill sets you must hire for. You're not going to train your own models -- most companies won't do that -- but your team must be familiar with the AI technology stack, harnesses, large language models and agentic AI. At the same time, they can't pull the trigger and just say, 'We're going to have a relationship with Anthropic' and hand it over like we're handing over our office suite to Microsoft. If you do that, then you're not owning your intelligence.
I heard someone say that intelligence will be a supply chain issue. How many businesses are subject to the price of oil? Not all of them. But we are all going to be subject to the price of intelligence. Your CIO must understand that they're the linchpin in their company's supply chain, even if it's a knowledge-based company. So, that must be sped up and slowed down in some places, and they have to think about that carefully.
First, most CIOs are resistant, in my experience, to using Claude because it's yet another tool. Claude's way better than Copilot. Use it.
Most CIOs want fewer vendors and projects, because it makes their lives easier … but that's not the way you thrive.
Paul BaierCEO and co-founder, GAI Insights
Second, CEOs want CIOs to learn more. Doing pilots is easy. What's hard is getting things in production. CIOs must understand that although it's another project to manage, it will help the business.
Most CIOs want fewer vendors and projects because it makes their lives easier, which is true. But that's not the way you thrive if you're one of these industries that we call 'in the crucible,' really going through this change.
-- Paul Baier, CEO and co-founder, GAI Insights
Be liberal with experimentation and strict with governance. Move quickly on training people and on automating repetitive internal work, with a human reviewing each result. Once the access rules are in place, let employees build their own automations, too.
Be cautious with anything large, outward-facing or irreversible, like an agent emailing a client, moving money or publishing something.
Akhil VergheseCo-founder and CEO, Krazimo
Be cautious with anything large, outward-facing or irreversible, like an agent emailing a client, moving money or publishing something. Those should come to a person as a proposal, with the reasoning attached, and the person approves or rejects it. And set your success metric before you deploy, because teams bend it afterward to make the AI look good.
Basically, set up the best practices, the data labeling and the governance structure and let your employees figure the rest out. Gatekeep productionization, but not experimentation.
-- Akhil Verghese, co-founder and CEO, Krazimo
I want my tech leadership to move quickly on using AI tools -- whether that's mathematical optimization, generative AI or a combination of both -- to drive efficiency and strengthen their decision-making.
That said, I'd be cautious about allowing general-purpose AI models or generative AI solutions to make consequential decisions on their own. Critical decisions involve hard constraints such as budgets, capacity, regulations, resource availability and competing objectives. You need a technology that is built for that type of challenge. This is where optimization can complement other AI tools -- by providing rigor, transparency and explainability to help organizations avoid hallucinations and reach decisions that leaders can confidently justify.
-- Bruce Zicari, CEO and managing partner, The Bonadio Group
It's a delicate balancing act. CEOs want their CIOs to capitalize on the speed AI can bring to their organizations. An automation that used to take a statement of work and a quarter to build now takes a sentence and an afternoon. This means tedious, repetitive work has never been easier to automate. But speed still needs to come with judgment.
Anyone can use AI to build an automation now, but deciding what to automate inside a company with real people, real politics and a hundred decisions nobody wrote down takes someone who understands how the business actually runs. For instance, an access request can touch HR, legal and IT at once, so the person building it must see the whole system and set controls that match the level of risk.
We've actually seen a new job created to fill this need -- the automation engineer. The role brings speed and judgment together. The CIOs who are winning are hiring automation engineers to embed with teams, learn how the work really happens and build automations that fit the company as a whole.
-- Alex McLeod, co-founder and CTO, Serval
I want the CIO to move quickly around deployable and scalable business use cases. That requires working across product, transformation and business teams to take ideas from solution development and technical architecture through to services that can be deployed rapidly.
I also want speed around workforce transformation. Organizations need to retrain employees and redesign internal roles to enable effective human-AI collaboration.
Where I want more caution is in balancing innovation and security. We should foster fast AI experimentation while maintaining rigorous controls around data privacy, regulatory compliance, cybersecurity, accuracy and governance. The CIO must help the organization move quickly while maintaining the trust and transparency required across the technology stack.
-- Krishna Nacha, CEO, Integreon
Tim Murphy is a reporter covering IT strategy for Informa TechTarget, with a focus on IT leadership, governance, workforce skills and AI strategy. His enterprise technology coverage has earned multiple Azbee Awards.