Context
A $74M global market leader in underground construction technology. Products shipped to 100+ export-controlled countries across 14 market groups. Hardware-first culture. No AI capability. No digital platform. No product operating model. The company’s core product line had not seen a new flagship in over a decade.
I joined as Head of Product, reporting directly to the CEO with portfolio P&L authority. The mandate was a turnaround: kill misaligned investments, redirect capital, ship new products, build recurring digital revenue, and establish the product organization from scratch.
Challenge
The company’s field operators work underground. GPS-denied environments. Intermittent or zero connectivity. Extreme conditions. The existing workflow required operators to interpret raw frequency data manually, a skill that took years to develop. Skilled labor shortages were compressing the talent pool. Fewer experienced operators. More complex jobs. The same analog interpretation process.
A cloud-based AI solution was not viable. The environments where the product creates the most value are the environments with the least reliable connectivity. The AI had to live on the device itself.
Approach
I drove the product strategy for an edge-first AI platform with an on-board neural network. The architecture processed 8,000+ sensor data points per workflow into autonomous decisions at the edge without cloud dependency. The device updates its intelligence model periodically from the cloud. It acts independently in the field.
The development approach broke from the company’s engineering culture. I proposed and led the company’s first-ever field trials with customers, validating the AI against real operating conditions before committing to production. Testing with real operators in real ground conditions. A first for a company that had shipped products for over 30 years.
The AI platform was not a feature bolted onto an existing product. It was a new product architecture built from the ground up as part of the first all-new platform in nearly two decades.
Outcome
40% projected support call deflection. The AI automates the frequency interpretation that previously required a human expert. The $15.9M in identified lifecycle cost savings informed the portfolio investment strategy and the product line sunset decisions that freed capital for this program.
The platform launched at the industry’s largest trade show one month after my departure, unveiled by the company’s second-generation leadership. The product operating model, the team, and the roadmap governance I built are still in place.
Lesson
Edge-first is the only viable architecture for industrial AI in field environments. The temptation is to build a cloud AI product and assume connectivity. In construction, manufacturing, and logistics, connectivity is the variable, not the constant. The intelligence must live where the action happens. Cloud handles model updates, fleet analytics, and pattern recognition at scale. The edge handles decisions.
The second lesson: field trials are non-negotiable. The only way to validate an AI product in a physical environment is to put it in the physical environment. Simulation and test benches are not enough. I had to fight for the company’s first field trials. The data they produced justified the entire investment.