Wolf&Grayson EditorialArtificial intelligence has become one of the most powerful levers for enterprise transformation, yet the question many leaders still wrestle with is how to actually bring AI into an existing analytics ecosystem in a way that delivers measurable value. “AI generates ROI in three main ways,” says Ahmad Fattahi, Senior Director of Data Science at Cloud Software Group. “It improves productivity, reduces risk, and creates entirely new revenue streams.”
One of the industry’s most pragmatic voices on applied machine learning, he believes AI must first address real business challenges before it can scale meaningfully. AI’s value is a balance of productivity gains, risk reduction, and new revenue opportunities, all grounded in practical needs and strengthened by a trust-driven evolution shaped by better tools, human oversight, and shifting industry boundaries.
Productivity as an Engine for Transformation
With two decades of experience blending engineering depth with business strategy, Fattahi has watched the evolution of enterprise analytics unfold up close. He points to software development, where AI is reshaping work. Developers at every level once began projects from scratch, rewrote boilerplate code, and spent hours on repetitive tasks and documentation. “There was no discrimination in the process,” he says. “Highly experienced engineers were still spending time on non-critical parts of development.”
Generative AI changed the rhythm of that story. Suddenly, models could draft code, evaluate it, assist with reviews, and accelerate iteration cycles. The work that once forced software developers to wade through repetition shifted toward creativity and problem-solving. And for Fattahi, this chapter is only the beginning. The same pattern he sees in software development is now unfolding across the enterprise, where AI is steadily lifting the repetitive layers that once defined day-to-day work.
Risk Reduction in High-Stakes Environments
That momentum is especially clear in highly regulated sectors, where the stakes of getting routine work right are significantly higher. In sectors such as finance and pharmaceuticals, regulatory compliance is a costly and error-prone necessity. Before the introduction of AI-driven workflows, organizations often relied on large teams to prepare filings and conduct evaluations. Human reviewers were essential, but human error was ever-present.
Fattahi recalls one case where a financial organization reduced the cost of a regulatory process from over $500,000 per year to only a few thousand simply by implementing generative AI tools. “You reduce risk at the same time that you lower cost,” he says. “That dual outcome wasn’t possible before.” This rebalancing of efficiency and accuracy is why AI-enabled risk reduction is becoming a core priority across regulated industries.
Creating New Value with AI-Enhanced Analytics
Beyond efficiency, AI is now reshaping enterprise analytics platforms at their foundation. Traditionally, users needed to understand where to click, how to script, and how to connect visualizations to perform advanced analysis. AI has begun to remove these barriers. “Imagine expressing your analytical intent in natural language,” Fattahi says. “The generative layer handles the rest.”
From automatically generating visuals to writing scripts and assembling reports, AI transforms analytics tools into collaborative engines rather than complex interfaces. For vendors, this means entirely new product tiers and revenue models. For users, it means faster insights and lower technical barrier.
Fattahi’s own work in launching Spotfire Copilot reflects this shift. He has built AI capabilities that sit inside analytics platforms, not beside them, making intelligence a native part of the workflow.
Building Trust in the Age of AI Decisions
However, even with all the progress, accuracy concerns remain. AI hallucinations can erode user trust, especially when organizations must validate results manually. “If employees spend as much time reviewing AI outputs as doing the work themselves, the value disappears,” he says.
Many tools already exist to significantly improve reliability. Retrieval-augmented generation strengthens contextual grounding. Model-judging systems evaluate responses before delivery. Fine-tuned models support domain-specific accuracy. And human-in-the-loop designs ensure that AI pauses when confidence levels drop. “AI shouldn’t be compared to a universe with infinite perfect humans working 24/7,” he says. “Compare it to what is practically achievable with your actual resources.” In many environments, even imperfect AI outperforms constrained human capacity.
A Moving Boundary Between Humans and Machines
The question of AI replacing human experts often emerges in specialized fields such as chip manufacturing. Fattahi sees this as a conversation about trends, not snapshots. Two decades ago, drivers printed MapQuest directions and manually verified every turn. Today, most people trust navigation systems without hesitation. “The line keeps shifting,” he says. “More decisions move into AI’s domain every year.”
He cites the 250,000 fully autonomous weekly trips performed by Waymo as proof that high-stakes automation is already here. If the market commits to solving a complex problem, he argues, AI can and will advance to meet it.
“It’s not perfect, and it’s not useless,” he says. “The truth is almost always somewhere in the middle.” The key is identifying where AI already delivers value while keeping an eye on the expanding frontier of what it will soon be capable of.
To follow more of Ahmad Fattahi’s insights on AI, data science, and enterprise innovation, connect with him on LinkedIn or visit his website.
