Most companies implement AI to automate tasks. Ahmad Fattahi has spent 15 years proving that AI’s real value comes from eliminating work that shouldn’t exist in the first place. As Senior Director of Data Science at Cloud Software Group, he’s led teams that turned data into measurable value.
The problem with most AI implementations isn’t the technology but the approach. Organizations apply AI to marginal decisions instead of high-leverage operations. They build one-off projects that deliver short-term wins without creating scalable tools.
Making operations leaner, smarter, and faster requires automating invisible inefficiencies, focusing AI on decisions with the biggest levers, and building reusable tools instead of one-off projects.
Automate What You Can’t See
Inefficiencies often hide in plain sight as manual workflows, redundant tasks, or repeated low-value decisions that teams accept as normal. Invisible inefficiencies exist because teams stop questioning why work happens the way it does.
At Cisco, Fattahi’s team replaced a 40-person triage operation with an NLP-powered automation pipeline. “Same outcome, but 87% less effort,” Fattahi notes. “AI lets you spot that invisible part of your operation and then eliminate it.”
The 40-person team processed incoming cases, categorized them, routed them to appropriate teams, and tracked resolution. The work was necessary but repetitive, following patterns that machine learning could identify and replicate. Automating the pattern recognition and routing logic achieved the same triage outcomes with a fraction of the manual effort.
Most organizations see this team and think about making it more efficient. Fattahi’s approach questioned whether the team needed to exist at all. That’s the difference between optimizing processes and eliminating unnecessary work.
Focus on Where the Value Is
AI isn’t just for predictions but for optimizing the decisions that matter most.
“In the energy sector, our machine learning models fine-tuned drilling operations, thereby saving hundreds of thousands of dollars per engagement,” Fattahi explains. “The bigger the lever, the bigger the impact.”
Drilling operations involve numerous decisions about depth, angle, pressure, and timing. Small improvements in these decisions compound across operations, turning marginal gains into substantial cost savings.
Most organizations spread AI efforts across many low-impact applications instead of concentrating on high-leverage decisions. They automate administrative tasks while ignoring optimization opportunities in core operations that drive revenue or costs.
Automating administrative tasks saves hours. Optimizing high-stakes operational decisions saves hundreds of thousands of dollars.
Fattahi’s approach prioritizes AI deployment based on decision leverage. High-stakes, high-frequency decisions with measurable financial impact should be optimized first.
Build Reusable Products, Not Just One-Off Projects
Savings only scale when AI becomes a product that multiple teams can use instead of a custom solution solving one problem.
At Spotfire, Fattahi’s team built dashboards and GenAI-based assistants that let non-technical teams self-serve insights. “AI multiplied its impact as a result because it was usable, reusable, and accessible to everybody,” he notes.
The difference between a project and a product is reusability. A project solves one problem for one team and requires data science involvement for any modification. A product solves a class of problems for multiple teams and enables self-service without ongoing data science support.
Most data science teams default to projects because they’re faster to ship. But projects don’t scale. Every new request requires rebuilding a similar analysis from scratch. Tools scale because the same infrastructure serves multiple use cases without proportional increases in data science effort.
Waste Less, Win More
After 15 years of leading teams that turn data into value, Fattahi knows that automating invisible inefficiencies can unleash significant business value.
“Cutting costs isn’t about corners. It’s about cutting friction,” Fattahi concludes. “Don’t just report yesterday’s numbers. Use AI to shape tomorrow’s operations. Turn data into action and finally action into savings.”
Getting real value from AI doesn’t mean you need the most models or the biggest data science teams. You just need to eliminate unnecessary work, optimize decisions that move numbers, and build tools that compound impact.
Connect with Ahmad Fattahi on LinkedIn for insights on AI-driven analytics and operational efficiency.
