Why 95% of enterprise AI investments fail, and how Margot Duek built the clinical protocol the industry didn’t have.
When a client tells Margot Duek that “AI isn’t working” inside their company, she doesn’t reach for a tool, a model, or a roadmap. She reaches for a question.
“What does the process look like before AI?”
Most of the time, no one in the room can answer it. Not the CEO, not the operations lead, not the team that runs the workflow every day. The process lives in someone’s head. The decisions are improvised. The exceptions outnumber the rules. And the AI someone bought, hoping to fix it, is now sitting on top of all of that, accelerating chaos instead of resolving it.
This is the pattern Duek has seen in dozens of companies in Mexico, the United States, and across Latin America. It is also the pattern MIT confirmed at scale. According to the State of AI in Business 2025 report from MIT NANDA, 95% of enterprises that invest in artificial intelligence see no measurable return. The 5% that do share one trait: they work with external partners who diagnose the operation first and prescribe the system second.
Duek built TRiAGE Business Health to be that kind of partner.
A Pattern She Recognized in Her Own Work
Duek is not new to the technology. A computer engineer from Tec de Monterrey with an MBA, she programmed neural networks before the AI boom and worked on technical teams at PayPal and Twist Bioscience. After nearly a decade of building digital systems for businesses across Mexico and the United States, she watched the same failure pattern repeat in early AI engagements, and decided the industry needed a different protocol.
TRiAGE was born out of that failure pattern, not despite it.
“The first AI systems we built didn’t fail because the technology was bad,” she says. “They failed because no one had a clear picture of how the business actually worked underneath. Processes lived in people’s heads. Workflows had never been documented. We built on assumptions, and AI built on assumptions doesn’t fix the mess. It makes it more expensive.”
That experience cost months, money, and clients who walked away skeptical of a technology that, applied correctly, can transform an operation. Duek wanted a protocol that could prevent it from happening again. She didn’t find one. So she built one.
A Clinical Protocol for AI Implementation
TRiAGE is structured deliberately around medical metaphor, not for branding, but because the parallel is exact. A good doctor never operates without imaging, history, and diagnosis. A good AI partner shouldn’t either.
The protocol begins with the X-Ray, a 45-minute clinical reading of an operation, free of charge for qualifying companies. It is not a sales call. It is a diagnostic session that ends with a written assessment of where the business is bleeding time, money, or both.
If the company chooses to continue, the next phase is the Blueprint, a two-to-three-week deep audit involving interviews with the team, mapping of the operational anatomy, and identification of the actual bottlenecks. The output is a clinical record of the business that names what is broken, what should be built, and in what order. Without it, no system is built.
Only then does construction begin. Each engagement also produces what TRiAGE calls the Operational Anatomy, a document that explains, step by step, how each automated process works and how the team can sustain it without depending on TRiAGE indefinitely. Without that document, Duek argues, the system lives in the developer’s head, which simply moves the original problem one level up.
Engagements completed under the protocol have moved client intake cycles in financial services from five days to eighteen hours, cut CRM data entry in print operations from two hours to thirty minutes per person, and turned manually-written shift reports in manufacturing into real-time dashboards that surface errors in minutes instead of the next day.
The Hippocratic Guarantee
The clearest signal that TRiAGE is built around the patient and not the sale is what happens after delivery.
Duek calls it the Hippocratic Guarantee, after the medical principle of primum non nocere, first, do no harm. If the system does not produce the results defined in the Blueprint within the first 90 days, the engagement does not close. TRiAGE keeps working at no additional cost until it does.
It is not a marketing line. It is the contractual posture of the firm. “If we can’t help you, we tell you,” Duek says. “If a tool you can buy off the shelf solves your problem, we tell you that too. We don’t sell shelves of services. We prescribe what your operation actually needs, and we hold the clinical risk for it.”
That posture is uncommon in a market where most AI consultancies charge for advisory decks and ongoing retainers without ever taking ownership of outcomes. It is also why TRiAGE has, since launch, declined engagements that would have generated short-term revenue but no measurable improvement for the client.
A Different Standard in a Loud Market
Duek’s authority on the topic is not theoretical. She is a recurring speaker at Stripe and Tec de Monterrey, an invited guest at Platzi, ISDI Business School, and Lazos Internacional, the former National President of the Marketing Branch at CANACINTRA, and the leader of Stripe’s Community Builders chapter in Mexico City. She co-founded two Google Developer Groups during her university years, one of them, WonderCoders, focused on bringing more women into programming.
But the company she is most insistent about is TRiAGE, and the reason, she says, is structural.
“The agencies and consultants who promise AI transformation without a diagnosis are the reason the 95% number exists,” she says. “We’re not trying to sell more AI. We’re trying to make sure the AI that gets built actually works, and that the team can run it after we leave.”
In a market that monetizes promise, that distinction is uncomfortable. It is also, increasingly, what serious operators want.
Companies interested in beginning the diagnostic process can request an X-Ray at TRiAGE Business Health.
About Margot Duek
Margot Duek is the founder of TRiAGE Business Health, an AI consultancy that diagnoses business operations before prescribing AI systems. She is also the founder of Al Chile Media, a digital agency with more than 300 projects delivered across Mexico and the United States, and a recurring speaker at Stripe, Tec de Monterrey, and ISDI Business School.
Learn more: TRiAGE Business Health · Al Chile Media · YouTube · LinkedIn
