This interview analysis aquanto And it was written, edited and published according to our regulations. Emerj Sponsored Content Guidelines. Learn more about our thought leadership and content creation services. Emerj Media Services Page.
Service organizations that support energy, infrastructure, and data center assets are facing a structural mismatch. While uptime requirements have tightened to near zero, maintenance models and technician capabilities have not kept up.
Demand is accelerating fastest at the grid edge. The U.S. Department of Energy cites an analysis from the Electric Power Research Institute that shows data centers could consume up to 9% of U.S. electricity generation by 2030, more than double their share in 2023. That pressure is already reflected in the cost of downtime, with the Supply Management Institute reporting that unplanned downtime now costs the world’s 500 largest companies $1.4 trillion annually, or 11% of revenue.
On the other hand, less labor is needed to prevent downtime. The U.S. Bureau of Labor Statistics projects that the number of electrician job openings will increase by 81,000 annually through 2034, primarily due to retirees rather than new entrants.
Fragmented equipment data further widens the gap. The National Institute of Standards and Technology found that inadequate interoperability of facility and equipment data costs owners and operators of U.S. capital facilities $10.6 billion annually in operations and maintenance alone. This same fragmentation prevents technicians from providing first-time fixes without asset history, documentation, or suboptimal maintenance guidance at the time of service.
Without guidance on suboptimal maintenance, technicians alone cannot bridge this gap.
Emerj’s Yolandi de Weerdt appeared on the AI in Business podcast with Joe Lang, VP of Service Technology and Innovation at Comfort Systems USA, to delve into how AI is reshaping service operations in mission-critical infrastructure.
This article examines operational and strategic insights from Joe Lang’s perspective on AI-powered service transformation.
- Anomaly detection for condition-based equipment maintenance: Capturing real-time sensor data allows service teams to intervene based on equipment behavior rather than scheduled tasks, supporting a near-zero downtime environment.
- Prescriptive guidance for consistent technician performance – Integrating diagnostic evidence with real-time next-best-action recommendations helps technicians move from predicting failures to fixing them quickly.
- Operational transformation required to change maintenance workflows: By treating workflow changes as dedicated operational initiatives with clear ownership, committed resources, and continuous improvement, organizations can achieve reduced downtime and costs.
Listen to the full episode below.
Episode: How AI is reshaping service operations for mission-critical infrastructure – co-authored with Joe Lang of Comfort Systems USA
guest: Joe Lang, Comfort Systems USA Vice President of Services Technology and Innovation
Expertise: Service technology, AI-enabled service operations, field service innovation, customer experience
Easy recognition: Joe Lang is Vice President of Service Technology and Innovation at Comfort Systems USA, where he leads technology and innovation efforts focused on improving field service operations and the customer experience. He has held executive service leadership roles at Comfort Systems USA for over 18 years. Previously, he held leadership positions at Johnson Controls and York International, where he oversaw service operations and business growth. Lang also serves on the advisory boards of Field Service USA, The Service Council, and Aquant. He holds a bachelor’s degree in industrial technology from Purdue University.
Anomaly detection for condition-based equipment maintenance
Although maintenance cycles are intended to be at predictable intervals, equipment behavior often changes during the unmonitored period between maintenance cycles. If these anomalies are left unaddressed, they can lead to progressive equipment failure and avoidable downtime. The gap between planned maintenance and real-time reality is the operational risk that Joe Lang focuses on, and why he argues that organizations must treat real-time behavior as a source of truth.
What sets him apart in this episode is that anomaly detection is not an advanced AI capability, but rather an initial operational discipline that allows teams to address emerging problems before they develop into failures.
Lang emphasizes that companies are already collecting the sensor data they need. What is missing is the rigor to uncover deviations early enough for technicians to intervene. He argues that most failures are not surprising, but rather detectable variations in behavior that organizations simply aren’t addressing. For him, anomaly detection is the moment service work becomes proactive rather than reactive.
Lang describes operational risk as follows:
AI gives engineers a head start. When a system flags a deviation, it is often the earliest indication that something is deviating from normal behavior. By acting in the moment, you can prevent failure instead of reacting to it. It changes the rhythm of service work. Teams stop chasing emergencies and start addressing problems before they become critical.
— Joe Lang, Vice President of Service Technology and Innovation, Comfort Systems USA
Joe identifies operational requirements for implementing anomaly detection.
- Equipment priority assets: Deploy and validate sensors on equipment that creates the highest operational or contractual risk of downtime.
- Define deviation threshold: Establish clear behavioral triggers (temperature, vibration, pressure, load) that signal actionable drift rather than noise.
- Automate technician routing: Enable deviations to generate service tasks instantly without manual review or batch processing.
- Measurement of intervention timing: Track how quickly teams respond to deviations and correlate early intervention with fewer avoided failures and fewer emergency dispatches.
These capabilities establish what Lang advocates as an operating floor: a service organization that responds to real-time behavior rather than scheduled assumptions.
Prescriptive guidance for consistent technician performance
Lang makes a clear distinction between predictive maintenance and prescriptive maintenance, the transition from predicting failure to recommending the next best action. He notes that many organizations still replace components such as air filters on a set schedule even when pressure drop data indicates clean operation, and that aligning interventions to actual equipment operation rather than calendar assumptions fills gaps in prescriptive guidance.
Technician performance fluctuates when faced with unfamiliar equipment, ambiguous symptoms, or failure modes that exhibit different symptoms each time. Lang’s point is that this disparity is not a personnel problem, but an information problem. Technicians start from different baselines because the diagnostic evidence they need is scattered across disconnected systems, buried in PDFs, or locked away in their personal experiences.
Prescriptive guidance stabilizes variation by giving all technicians the same informed starting point. When service history, OEM documentation, resolution patterns, and equipment context are integrated and delivered in real-time, the system can reveal not only the likely failure but the next best action to resolve it before the panel even opens. Technicians still make calls, but instead of guessing, they start with diagnostic information that the organization has accumulated.
Lang emphasizes that this does not replace the judgment of engineers. It’s about removing the uncertainty in the first 10 minutes that leads to inconsistent results. When the probability of failure is high and recommended interventions arrive at the start of service, diagnostic breadth is narrowed, first-time fix results improve, and constrained employees operate at a more stable baseline of performance.
Sources of evidence that Joe identified as prescriptive guidance:
- Service history: Actual failure modes, symptoms, and corrective actions that show how the problem actually occurred and how it was resolved.
- Manufacturer documentation: OEM guidance that defines intended behavior, known failure paths, and validated diagnostic procedures.
- Solution pattern: Proven fixes that consistently solve problems in the field reveal reliable, proven interventions.
- Equipment context: ID, configuration, and operating conditions to ensure that recommendations reflect the actual behavior of the asset.
For these sources of evidence to be available at the moment of service, Lang emphasizes that they must exist within a single structured data environment that models can reason about and provide to technicians in real time.
Joe identifies operational data workflow design choices to make prescriptive guidance effective.
- Integrate diagnostic evidence: Incorporating service history, OEM documentation, and resolution patterns into one structured environment allows models to reason across the complete diagnostic evidence set.
- Provide guidance in real time: Reveal likely failures and next-best actions directly in a technician’s workflow, such as a mobile app, work order, or dispatch interface.
- Anchor recommendations to equipment context. Ensure that guidance reflects the identity, composition, and behavioral history of specific units, rather than relying on general assumptions.
These diagnostic workflow design choices create the consistent starting point that technicians need. That means a real-time, evidence-based baseline that reduces variability and enhances first-fix performance across a constrained workforce.
Business transformation required to change maintenance workflow
Lang emphasizes that modernization begins with identifying, categorizing, and organizing assets into logical equipment groups. Without structured asset data, centrally managed documentation, and service history, organizations cannot reliably apply anomaly detection and prescriptive guidance.
Lang explains that this is because organizations tend to treat migration as a part-time endeavor, rather than dedicating resources to it as a dedicated effort with clear ownership, by getting batch data here and loading it into a platform there.
He argues that this lack of resources is the most common reason modernization efforts stall. The team spent a year building the data infrastructure, but saw no tangible results because they didn’t have the people in place to make the project a success.
Lang frames operational reality as follows:
This is where you change planes mid-flight. It must be modified so that it can continue to fly, land, and take off again. You need to resource this correctly when you start down this path.
— Joe Lang, Vice President of Service Technology and Innovation, Comfort Systems USA
Joe identifies operational requirements to treat maintenance workflow changes as a dedicated effort.
- Assign exclusive ownership. Have a specific team or project leader on staff responsible for the transition, rather than distributing the work to existing roles as a secondary task.
- First, inventory and categorize your assets. Group equipment by type and component similarity before applying anomaly detection or prescriptive models across your fleet.
- Centrally manage manuals and service history: Integrating OEM documentation and equipment records into a database allows technicians to access information directly in the field rather than having to search for it upon arrival.
- Ensure sufficient resources. Treat this effort as a fully staffed program rather than an incremental data project layered on top of your existing workload.
For Lang, the combination of committed resources and structured asset data determines whether an organization actually reduces downtime and costs or simply accumulates unused data infrastructure.
