Ask most people what gives AI its edge and they'll likely point to speed, automation or the aura of generative AI tools.
But according to experts at SAS Innovate 2025, AI's real competitive advantage isn’t the algorithm – it’s the ability to use it responsibly to make trusted, faster, better decisions.
As SAS approaches its 50th anniversary, Chief Technology Officer Bryan Harris used the milestone not to look back but to lay out a bold path forward: Building smarter, trusted systems.
Here are five key takeaways from SAS Innovate’s opening session with Harris – packed with real-world examples and quotes that challenged conventional thinking about AI’s value.
1. AI’s superpower is decision intelligence
Over the years, we’ve seen waves of technological breakthroughs: First came the PC and internet, then the cloud, machine learning and deep learning. Now, it’s all about AI – in every flavor imaginable: generative AI, agentic AI and quantum AI. The rapid pace of innovation can feel overwhelming, even to the companies building it.
“This pace of innovation can feel intense and overwhelming, even for the companies that create it,” Harris said. “But every subset of AI is fueling one thing: decision intelligence.”
The real value of AI lies not in what it creates, but in the decisions it enables.
You’re not looking for just any decision. You need a decision that delivers outcomes that outperform. You need a decision that helps you compete to win in the market. You need a decision advantage. Bryan Harris, SAS CTO
2. Generative AI won’t fix broken business models
Generative AI is making waves in areas like customer service and text summarization, but it’s not a magic fix for all business problems. Harris called out the myth that simply adding generative AI to a process will solve all your issues.
“It feels like all you need to do is sprinkle a little generative AI in the enterprise and all of your problems will magically disappear,” he said during his keynote.
While generative AI has impressive applications, it also has serious limitations, especially when not applied properly.
“You can have a good, unbiased AI model – but feed it the wrong data and you have a biased outcome,” Harris warned.
In fact, a recent study revealed that large language models tend to recommend higher interest rates and loan denials for Black applicants compared to white applicants, even when credit scores are the same. On average, Black applicants need credit scores 120 points higher to get approved at the same rate, according to the study.
Harris emphasized that such discrepancies are systemic across major LLMs. A simple solution, like adding more emphasis to the prompt, doesn’t solve the issue.
“That’s why LLMs and prompt engineering alone are not enough for most enterprise use cases,” he said. “You need an orchestration of LLMs, machine learning, APIs and more to ensure accuracy, fairness and governance.”
This points to the shift in industry focus from generative AI to agentic AI – moving beyond just creating content to making AI-driven decisions that are fair, accurate and effective.
3. From generative AI hype to agentic AI reality
While the tech world still loves generative AI, SAS is already deploying what’s next: agentic AI – systems that reason, act and collaborate with or without human input.
And it’s moving fast. The global agentic AI market is expected to exceed $70 billion by 2030, with 52% of organizations adding AI agents to workflows this year.
So, what does SAS mean by agentic AI?
“SAS thinks of it as a spectrum between ‘human out of the loop’ and ‘human in the loop,’” Harris said.
“Bottom line … we don’t just give you an agent – we give you transparency and explainability.”
At one end, agents act autonomously, like denying fraudulent transactions in real time. On the other hand, they flag irregular cases, provide risk scores and offer context for human review.
This is shared intelligence in action – already at work in SAS® Intelligent Decisioning. In a demo, Harris showed how SAS® Viya® can flag mortgage cases, explain its reasoning and give reviewers full transparency from model cards to decision lineage.
These agents aren’t limited to fraud or loans. SAS customers use them across industries. They are built in low-code environments and combine LLMs, rules, models and APIs into scalable, reusable workflows.
The result? Smarter, safer and more accountable AI with governance and fairness built in.
4. What if... digital twins could think like operators?
What if you could ask your factory floor, “What if?” and get answers backed by real data, simulation and AI?
That’s what SAS, Epic Games and Georgia-Pacific are building together.
At Georgia-Pacific’s Savannah River Mill, which is over 400 meters long (larger than four football fields), they built a digital twin to test how autonomous guided vehicles (AGVs) move through the facility. These machines are critical, but finding the right fleet size and routing is tough.
“Today’s manufacturing plants are complex environments with AGVs, robotics and people all working together,” Roshan Shah, VP of AI at Georgia-Pacific, explained. “When things go wrong, it gets expensive and people can get hurt.”
Using SAS Viya, the team ran simulations inside Unreal Engine to test AGV strategies. They assumed more AGVs would improve productivity, but the digital twin showed otherwise.
“Adding AGVs seems intuitive, right? More workers should mean more output,” Shah said. “But in Unreal, the cost of extra AGVs actually slowed things down.”
The optimal number turned out to be 47, boosting performance by 8%.
“Instead of building and testing these solutions in the real world,” said Bill Clifford, VP of Unreal Engine Ecosystem at Epic Games, “one click lets us find the best configuration among all possible worlds.”
SAS even added a “factory editor” to the simulation, letting managers adjust layouts, reroute AGVs, or toggle between real and synthetic data in a gamelike interface.
As Shah put it, “Imagine being a plant manager and being able to adjust operations in this manner. It’s a complete game-changer. Pun intended!”
This is not just visualization. This is decision making powered by AI and immersive tech.
Understanding digital twin technology
5. Transforming optimization with quantum AI
Harris pivoted to quantum AI as a game-changer for tackling large-scale optimization problems.
“Imagine having a digital twin that answers your most difficult 'what if' questions instantly,” he said, noting how quantum accelerates data processing from hours or days to minutes.
He referenced longtime collaborations with global organizations like Procter & Gamble. “We’ve worked together for decades to improve performance and reduce cost,” Harris said. “Now we’re exploring how quantum AI can take that to the next level.”
As Krista Comstock, Director of Digital Innovation at P&G, put it, “One of the most significant takeaways is the improvement in productivity and performance. Quantum, being a probabilistic problem solver, doesn't always give the best answer, despite being fast. Whereas traditional solvers would give higher quality answers, but take longer to solve. The hybrid approach gives us the best of both worlds: both speed and quality of the solution.”
From reducing compute times to boosting the efficiency of digital twins, hybrid-traditional quantum approaches are already showing real promise in driving transformation at scale.
The bottom line
AI may be evolving fast, but SAS is helping customers evolve faster, with tools that make every decision more intelligent, explainable and effective.
Whether it's through digital twins, agentic AI or quantum AI innovation, one thing’s certain: The future belongs to those who build trust into every layer of their technology.
