Twenty years of technology adoption at Microsoft and Cisco taught me one consistent truth that almost every AI conversation in sports skips over: the technology is rarely the hard part.
The hard part is change, and change is hardest when you can’t picture what’s on the other side of it.
I’ve been in enough rooms — with Fortune 500 companies, with government agencies, with organizations that had every reason to move fast — to know what stalls adoption. It’s not capability, it’s the moment someone at the table asks whether the investment is worth it and whether the disruption to existing workflows is justified. What they’re doing now feels known and the outcome is predictable, even if the process is slow or inefficient. AI asks them to give that up for something they can’t fully visualize yet, and that’s a new way of operating that exists only in a vendor’s deck and a demo that may or may not reflect reality.
That’s a hard ask. And in professional sports, where the cost of a wrong bet shows up on the scoreboard, it carries more weight than almost anywhere else I’ve seen.
The NFL Has Always Navigated This
Former NFL scouts and coaches who work at SūmerSports are quick to point out that this isn’t the first time the sport has faced a technology shift of this scale. When 16mm film gave way to beta tape, organizations had to fundamentally rethink how they studied the game. That transition is the literal origin of “look at the tape.” Video producers who once worked in dark rooms had to become IT-literate overnight, and resistance came at every turn, but every organization that made it through realized the transformation had propelled them forward.
AI is asking the same thing of football organizations today.
The sports executives and operations leaders I talk to have heard the AI pitch. Most have sat through the demos and what they haven’t seen is AI that works the way it was sold to them, and that gap is where the skepticism lives. Most have already lived through at least one initiative that promised everything and delivered friction. That doesn’t make them resistant to progress, it makes them careful.
The Real Architectural Problem
Here’s where it gets complicated.
To get real value from AI, it has to know your organization. A generic model gives generic answers. To get AI that understands your scheme, surfaces insights from your film, and speaks your language, your data has to be part of the system.
NFL front offices aren’t working with generic enterprise data; they are working with the most competitively sensitive information in the building like playbooks, depth charts, pre-draft intelligence, etc. The value of customization is clear, but the potential exposure isn’t acceptable. So nothing moves.
What makes this hard is that security and customization are pulling in opposite directions, and most AI deployments haven’t resolved that tension. That’s the real reason so many organizations are still running pilots.
The Answer: The Intelligence Comes to the Data
This is exactly what SūmerBrain on Cisco AI PODs is built to resolve.
SūmerBrain is SūmerSports’ AI engine, purpose-built for professional sports and validated by 500 years of combined NFL front office experience. It handles complex, multi-variable queries no general tool can answer. Cisco AI PODs are pre-integrated, NVIDIA-validated full-stack infrastructure that runs AI workloads entirely within an organization’s own environment. For a deeper look at the infrastructure architecture, Jeremy Foster, SVP/GM, Cisco Compute, breaks it down here.
Together, the data never moves. No cloud dependency and no black box because the intelligence comes to the data, not the other way around. That means an organization can get fully customized AI without any of it leaving the building.
The security and customization problem stops being a tradeoff. Both are solved in the same architecture, and when that blocker is removed, the path to ROI gets considerably shorter.
From Pilot to Production
Two decades of watching enterprise technology deployments taught me that organizations don’t fail at AI because the model underperforms. They fail because everything around the model — the workflows, the buy-in, the operational integration — is slower and harder than anyone planned. The pilot works, but the deployment doesn’t. That pattern has repeated itself across every major technology shift I’ve seen, and AI in sports is no different.
The organizations that moved from experimentation to execution weren’t the ones with the best technology. They were the ones that removed the friction between insight and action and trusted their people to use the tools.
Cisco AI PODs were built for exactly this. Pre-integrated and validated means teams aren’t assembling components and hoping the seams hold. The system is ready from day one, in their environment, at the reliability standards a professional sports organization requires.
The Change Management Payoff
What we’re announcing with Cisco isn’t a research partnership or a proof of concept. It’s a deployment model for sports organizations that are done experimenting and ready to get ROI.
Once the data security concern is resolved and the infrastructure is in place, something shifts. The AI can be customized to the organization. It knows the language of the building, and when the AI knows you, the gap between an insight and a decision shrinks to seconds. That’s where the return on investment actually lives; not in the technology itself, but in how fast a well-implemented system lets your people move.
That’s the change management payoff. After twenty years helping large organizations reach it, I can say with confidence: we’ve built the right path there.
