When Stake Center set out to improve efficiency and reduce excavation damages, the goal wasn’t to deploy artificial intelligence for the sake of innovation. The objective was far more practical: manage overwhelming locate volumes, prioritize risk, and protect critical infrastructure.
What followed was a year-long learning journey using Irth Insights — one that delivered measurable results, including a 20% increase in field locate efficiency and a reduction in excavation damages, but also surfaced important lessons the industry can learn from.
Stake Center shared what worked and what didn’t at User Summit 2025. This is their story, and other lessons discussed in the session.
The 20% improvement in locate efficiency did not come from simply clearing more tickets automatically. It came from reordering priorities around risk.
Stake Center knew risk must override clearing decisions. If a ticket crossed a defined risk threshold, it didn’t matter how “clearable” it appeared — boots on the ground were required.
Irth Insights identifies 50% of damages in the top 8% of tickets.
This knowledge shaped how the system was configured:
Key takeaway: AI should not replace judgment. Risk-first logic prevents efficiency gains from turning into safety liabilities.
Before implementation, human screeners spent valuable time reviewing both:
Irth Insights exposed the inefficiency of this approach.
By stacking risk and clearing models together, Stake Center learned to:
Lesson learned: The real efficiency gain was better allocation of human expertise.
One of the most important realizations was that scores alone don’t reduce risk.
A risk score on a ticket doesn’t prevent damage unless it:
Stake Center used risk scores to:
Lesson learned: AI delivers value when it is embedded into workflows that actively drive change in the field.
One of the early learnings was understanding that clearing and risk thresholds can’t be “set and forget” it.
Models retrain frequently as new data flows in, and risk and clearing behavior shift over time by state, by excavator, by season.
Stake Center learned the importance of:
Lesson learned: Successful AI programs require operational ownership. Someone must actively manage the system, not just deploy it.
Explainability becomes critical when there is skepticism about AI.
When users could see:
…confidence increased dramatically.
This transparency allowed teams to:
Lesson learned: Trust in AI comes from understanding.
If tickets were repeatedly “cleared” incorrectly by people in the field, for example, the model learns that behavior — sometimes to the frustration of excavators and operators.
This can be addressed by:
Lesson learned: Bad habits scale quickly if left unchecked. It comes down to consistent and accurate data collection.
Stake Center limited clearing automation to fiber tickets only.
By scoping one use case, they were able to:
Lesson learned: Start narrow. Prove value in a controlled environment before expanding.
A 20% efficiency gain changed the workforce equation.
Stake Center learned that:
Rather than increasing pressure, efficiency helped reduce turnover, one of the industry’s most persistent challenges.
Lesson learned: There are multiple metrics for success when implementing Irth Insights.
Stake Center’s experience reinforced a critical truth for the industry:
AI is inevitable — but results are not.
The gains came not from technology alone, but from:
For organizations considering AI-driven risk and clearing, the biggest lesson is this:
The value isn’t in automating decisions — it’s in making better ones, faster, with the right people involved at the right time.
That’s where the real efficiency — and safety — are realized.
Check out the full session
To better understand how Irth Insights can enable your team to reduce damages, improve safety, and increase efficiency, schedule a demo today.