The AI pilots that drive profits at tech firms

“The era of AI is not just about adopting cutting-edge technology. It's about transforming business models, strategies and operations.”

Katie MacQuivey 

Partner, Business Consulting
Grant Thornton Advisors LLC

Many leaders are trying to determine which AI projects they should scale by applying the same efficiency standards they use to evaluate RPA projects. While efficiencies can provide important savings, AI can do more. To achieve the lasting enterprise value AI promises, leaders need to help recognize, pilot and scale projects that align with its real potential. “The era of AI is not just about adopting cutting-edge technology,” said Grant Thornton Business Consulting Principal Katie MacQuivey. “It’s about transforming business models, strategies and operations.”

“Many AI pilots focus on operational efficiencies — cost savings, speed, and headcount reduction — but that’s a short-term view.” MacQuivey said. For long-term benefits, leaders should look through the lens of their vision and strategy. “Rather than pursuing workforce reductions, consider how you can optimize existing resources to enhance productivity and drive company growth. Enabling team members to operate at higher efficiency creates opportunity for greater returns.”

Greater returns can justify greater AI investment. Grant Thornton Technology Industry Leader Andrea Schulz said, "While substantial investment continues to flow into AI pilots, the critical question remains whether we'll see productivity gains that justify the expenditure. Initial implementations focused primarily on operational cost savings, but early adopters are now experiencing something more significant — accelerated revenue growth driven by improved product quality. The economic returns are beginning to manifest, suggesting we may be transitioning from cost-cutting exercises to genuine value creation."

Schulz suggested an example in vibe coding, where AI develops software based on natural-language inputs. “Engineers are becoming a lot more powerful and producing more. That can also come back to a build-it-or-buy-it decision for a lot of software solutions. I know companies that need bespoke software solutions and are empowering non-developers to build them internally rather than using external teams.”

So, how can tech leaders identify the right AI pilot projects to help drive long-term enterprise value?

Strategy, maturity and risk

“Right now, I think there’s a proliferation of people who want to say they’re using AI, and so it’s a hammer looking for a nail,” said Grant Thornton Technology Modernization Services Principal Zac Taylor. “Even if business processes don’t need AI to be effective, sometimes people try to make them AI-driven for marketing, publicity or other reasons.”

“Think about where it’s on-strategy to develop an AI solution,” Taylor said. “Where does it meet your business strategy and where is it on point for your technology strategy?” Once leaders identify strategic use cases, they can move to other considerations. “Then, it’s a risk and maturity conversation. Assessing risk and maturity means asking if the solution delivers reliable results without exposing your organization to risk—a factor mainly shaped by the quality of your backend data.”

“A lot of AI pilots have limited inputs, and therefore they’re getting limited returns … They aren’t enterprise solutions in that they don't tie into the general core workflows and data of an organization.”

Zac Taylor 

Partner, Technology Modernization Services
Grant Thornton Advisors LLC

“A lot of AI pilots have limited inputs, and therefore they’re getting limited returns,” Taylor said. “We’re seeing many innovations that are really point solutions. They aren’t enterprise solutions in that they don’t tie into the general core workflows and data of an organization. Certain things, like a chatbot, might be relatively simple. While such an approach may be intriguing, effective process enablement requires integrating appropriate models with relevant data. This alignment allows organizations to, for instance, identify customers suitable for new services or determine potential markets for expansion.

Companies know that they need to ensure security when integrating enterprise data, but Taylor said they might not anticipate the typical issues with data hygiene. “Most companies don’t have sufficient data hygiene to provide enterprise data to AI agents,” Taylor said. “If you can’t trust your system of record’s data because it’s not as clean as it should be, the value you get from running AI on top of that will be limited.”

MacQuivey added, “You have to put in the effort to make sure that the data is clean, and that it fits in with your other business processes.” Data hygiene is just one aspect of the enterprise preparation that is required for other technology initiatives but often forgotten for AI and especially GenAI.

“When pilots align with companies’ strategic intent and vision, change management becomes more seamless: employees understand the purpose, are motivated to adopt and upskill, and become champions of transformation.”

Katie MacQuivey 

Partner, Business Consulting
Grant Thornton Advisors LLC

To successfully scale AI pilot projects, companies need enterprise support. “Pilots often fail because they’re too narrowly focused or built in silos,” MacQuivey said. “AI pilots should be treated as strategic investments, not side projects. Organizations must identify and focus on priority AI initiatives to maximize realization of ROI.”

“Before scaling an AI pilot, it’s essential to prepare your organization — especially the people who will be impacted,” MacQuivey said. “When pilots align with a company’s strategic intent and vision, change management becomes more seamless: Employees understand the purpose, they are motivated to adopt and upskill, and they become champions of transformation.”

Taylor added, “Culture is a massive part of success with AI, and we’re all a bit excited — but also hesitant — about what its impacts could be.” MacQuivey said, “Employees often hesitate to embrace AI, whether that’s due to fears of job displacement, unfamiliarity with the technology, or a lack of relevant skills. Embedding AI into a broader strategic vision, helping people understand its purpose, and equipping them with the right tools and training can turn uncertainty into momentum and make adoption truly successful.”

MacQuivey outlined that, “This is where you need to align the executive team on strategic intent, then perform use case identification and prioritization based on a wide array of factors.” Those factors will continue to evolve, so decision-makers need to re-evaluate and reprioritize use cases regularly. She also highlights the need for agility: "Be ready to act when a new use case arises — you'll know what's needed and how to prioritize."

“Having clarity on where you’re going helps you guide the investments in pilots and ensure that the infrastructure and data are in place,” MacQuivey said. Leaders also need a strategic perspective to avoid exposing their firms to long-term risks in the pursuit of short-term gains.

“Some companies are so focused on the efficiency play right now that they could find they’ve created risks for themselves and lost subject matter expertise 5 to ten years into the future.”

Zac Taylor 

Partner, Technology Modernization Services
Grant Thornton Advisors LLC

While efficiency is an important goal for digital transformation, it can also distract leaders from balancing long-term resilience. “I think part of the risk, when organizations are myopically focused on efficiency right now, is that they’re going to lose something in the future without developers and engineers who know how to program on their own,” Taylor said. “Bots are becoming better and more efficient — no doubt that's true — but we need to be careful about how quickly we get rid of thinking in our environments. Some companies are so focused on the efficiency play right now that they could find they’ve created risks for themselves and lost subject matter expertise five to 10 years into the future.”