AI in Action: What We Learned from Our Latest Rocky Mountain Webinar
- May 9
- 5 min read
Recapping our May 7 panel with experts from healthcare AI, data governance, clean energy, and legal tech.
The Rocky Mountain French-American Chamber of Commerce brought together three innovators on May 7 for AI in Action: Real-World Applications Across Industries — a live panel that kept firmly to the present tense. No speculation, no crystal balls. Just concrete use cases, honest caveats, and practical guidance for business owners and professionals navigating one of the most consequential technology shifts of our time.
Moderated by Anna-Sophia Kristjansson, CEO of Lexicon Lens and a former FACC New York leader, the conversation featured Mathilde Fievez (Breakthrough Energy Fellow at Beyond Silicon), Franck Nanie (Founder of The Talisman Group), and Frank S. Giaoui (Founder of Optimalex and lecturer at ASU College of Law). Here is what stood out.
AI Is Not Just ChatGPT
Frank Giaoui opened the panel by resetting expectations. AI encompasses natural language processing, classical machine learning, predictive modeling, and the emerging wave of agentic systems — generative AI is just one slice of a much broader landscape. At its core, he explained, any AI system does the same thing: it learns from known data, recognizes patterns, and makes predictions on new scenarios. "It's not magic. It's probabilistic."
He distilled the practical use cases into two buckets that every business owner can understand. The first is automation — eliminating manual, repetitive tasks. At Optimalex, this looks like medical chronology: a process that once took a lawyer or doctor ten hours to complete manually now takes three minutes. The second is decision support — helping humans make faster, more accurate decisions. AI-assisted analysis of lung MRIs, for example, has demonstrated measurably better accuracy than manual review alone. In the insurance and legal world Giaoui knows well, every day of delay before the right decision costs approximately $1,000 per claim. Across hundreds or thousands of annual claims, that math adds up fast.
The Data Foundation Has to Come First
Franck Nanie brought a grounding counterpoint: AI is only as good as the data that feeds it. His fractional CDO practice exists precisely because companies are investing in AI capabilities before they have their data house in order. Garbage in, garbage out — as he put it.
Before launching any AI initiative, Nanie recommended a checklist approach: clarify your goal (automation or decision support?), classify your data and understand which of it you do not own, assess your regulatory environment, establish a governance baseline, define expected ROI, and assign clear ownership for accountability. That last point matters more than people expect. When something goes wrong, ownership is how you reverse-engineer the problem.
He also named a risk most organizations underestimate: shadow AI. Employees across industries are quietly feeding real business data into unauthorized tools with no data processing agreements and no recourse. "The exposure is real," he said. For any business that has already deployed AI without a DPA in place, his immediate advice was to inventory every tool in use (sanctioned or not), identify what data each one touches, and build a remediation plan from there.
In R&D, Start with the Bottleneck
Mathilde Fievez offered a different vantage point: a fast-moving startup manufacturing solar cells every week, operating at the intersection of aggressive timelines, limited resources, and high uncertainty. Her AI journey was not driven by strategy documents. It was driven by a team member who spent every weekend manually assembling operational reports — a bottleneck that only became hers when he left for another company.
Her solution was elegantly practical. Unable to feed proprietary process data into external models, she used ChatGPT with anonymized fake datasets to iteratively build a Python script — run locally in Jupyter Notebooks — that pulls data from a secure server, generates graphs from Excel files, resizes images, and compiles everything into a PowerPoint automatically. No coding background required. She taught it to her team.
The lesson she took from deployment surprised her: the human side was harder than the technical side. Young hires embraced the tool immediately. More experienced colleagues needed time, explanation, and reassurance about what the tool was actually doing and why. "You almost have to convince them to use it even when it's already working." Her advice to anyone else deploying: align your team on the workflow before you automate it. Different people have different ideas about how a process should work, and automating the wrong version of the workflow compounds the problem.
The Cost of Waiting Is Higher Than You Think
One of the most memorable segments of the panel was Giaoui's breakdown of opportunity costs for businesses that have not yet acted. A wellness practice managing 100 appointments per month loses 15 to 20 percent of revenue to no-shows; AI-powered automated reminders alone have demonstrated $6,000 in recovered revenue annually, with chatbot integration adding another $30,000 in impact — a nearly ten-to-one return on a roughly $5,000 investment. An online education platform with 250 students can achieve a seven-times ROI in year one simply by deploying a chatbot to handle the administrative questions that cause students to drop out. A ten-person consulting firm that reduces lead response time from 24 hours to one hour can see conversion rates improve from 15 to 20 percent — a compounding effect that can exceed $500,000 annually.
"The later you come in, the more you let your competitors gain on market share," Giaoui said. "When you are a small business, you cannot have the luxury of waiting."
Most Businesses Are Below the Regulatory Threshold
One of the most practically useful moments of the panel came when Giaoui addressed the governance question directly. Many attendees, he noted, are not in highly regulated sectors. If you are a B2B business with under $25 million in revenue, not managing more than 100,000 consumer records, and not making decisions that directly impact employment, credit, housing, or healthcare outcomes — and if you keep a human in the loop — you are largely below the radar of current AI regulation. "While you should always be mindful, the cost of not implementing AI or delaying is much higher."
That said, for those operating in cross-border environments, the picture is more complex. Fievez pointed to solar supply chains spanning Asia, the US, and Europe, where tariff regimes and regulatory frameworks interact in increasingly complicated ways. Giaoui noted the divergence between FDA processes and their European equivalents, and even between US states — California and Colorado are now among the strictest regulators in the country on AI-adjacent issues like insurance. His practical advice: when in doubt, comply with the strictest standard applicable to your market.
Where to Start: The Panel's Consensus
When the audience poll revealed that 50 percent of attendees did not know where to start, the three panelists converged on remarkably similar advice.
Identify your single biggest pain point — the place where you are losing time or money most visibly. Find one champion inside your organization, someone willing to own the initiative. Measure your baseline before you begin, so you can track progress. Start small, prove the model, and expand from there.
And when skepticism surfaces — your own or your colleagues' — Nanie offered a reframe: "When you ask AI to do something for you, you can also ask AI to demonstrate to you why this is the right way. This is how you gain control. AI is not ready to be on autopilot. It needs you, and it's going to make your life better."
Thank You to Our Panelists
We are grateful to Mathilde Fievez, Franck Nanie, and Frank S. Giaoui for bringing their experience and candor to this conversation. Each of them is a member of the Rocky Mountain French-American Chamber of Commerce, and each is doing meaningful work at the intersection of technology, business, and policy.
If you missed the live session or want to explore how AI and international business intersect for your organization, we invite you to connect with us and become a member.
The Rocky Mountain French-American Chamber of Commerce serves Colorado, Arizona, Utah, Wyoming, and New Mexico. To learn more about upcoming events and membership, visit rmfacc.org.
Comments