3 insights from Episode 15 with Cam Stevens

Episode 15 of Headline Podcast, published on March 18, 2026, features Cam Stevens, safety technologist and CEO of PKG, in conversation with Andreza Araújo and Dr. Megan Tranter.

His central argument changes how leaders should evaluate safety technology because a signal has no protective value until someone can interpret it, act within a defined window, and verify that the work changed.

What makes a technology signal actionable?

A technology-generated safety signal becomes actionable when it names a decision, an owner, and a response window rather than merely adding another number to a dashboard. Cam Stevens places the problem before the product, which means a site should know what exposure or decision gap it wants to change before it selects a tool.

That distinction matters for executives who receive dozens of alerts each day. A prediction score, a voice transcript, or a wearable notification may look precise, yet none of them tells a supervisor whether to stop work, change the plan, add a control, or wait for more evidence. The missing element is not always data. It is decision design. The related guide on technology adoption and psychosocial blind spots applies the same question to work design.

“If we're very clear on the problem to solve, then a technology catalog is excellent. If we're not clear on the problem to solve, a technology catalog is very dangerous.” On Headline Podcast, Cam Stevens connected technology selection to problem definition and decision quality.

OSHA describes safety management through leadership, worker participation, hazard identification, hazard prevention, and program evaluation in its safety and health program guidance. That structure supports Cam Stevens's point because technology should strengthen an existing management decision instead of disguising the absence of one.

As Andreza Araújo argues in Safety Culture: From Theory to Practice, safety culture is visible in the decisions that leaders reinforce under operational pressure. A digital signal therefore deserves the same test as any other control. What decision does it improve, and what happens when the signal is wrong?

Why ownership matters more than prediction

Signal ownership matters more than prediction accuracy when a safety decision must be made before the next shift. A tool can identify a pattern, but only a named person can decide whether the pattern requires escalation, verification, redesign, or no action.

Cam Stevens warned on the podcast that a technology catalog becomes dangerous when leaders have not defined the problem to solve. The same warning applies to artificial intelligence and predictive analytics. A model can produce a probability in less than 1 minute, while the organization still has no agreement about who reviews the result or what evidence is sufficient.

A practical governance brief should record 4 items. It should identify the signal, the accountable role, the response clock, and the proof that the field condition improved. If the response owner is a committee, the signal will often age before anyone accepts responsibility for it. Leaders can compare this with the decision-rights test for control ownership.

Andreza Araújo's work across 25+ years in executive EHS reinforces the need for visible accountability. The technology does not replace leadership presence. It gives leaders one more source of evidence, which remains weak until a supervisor or manager can connect it to a real work decision.

How new technology changes indicator meaning

New technology changes the meaning of an indicator because it changes the work system that produces the measurement. A lower count of reported issues may reflect better conditions, reduced attention, alert fatigue, fear of monitoring, or a reporting workflow that has become too slow.

Cam Stevens also described technology use as a source of changing psychosocial risk. On Headline Podcast, he said, “The changing shift in risk profile will be overwhelmingly psychosocial, driven by technology usage in our organizations.” That idea belongs in indicator governance because workload, autonomy, monitoring pressure, communication, and recovery can shift before a lagging injury measure moves. A tool that improves visibility while increasing friction may produce a cleaner dashboard and a less usable operation.

NIOSH explains through its Total Worker Health approach that worker protection and well-being should be considered together. For a technology pilot, that means leaders should track both the target risk and the work conditions created by the intervention.

The measurement question is therefore broader than whether the alert fired. Leaders should ask whether the worker understood the signal, whether the supervisor could respond, whether the response reduced exposure, and whether the new process created extra administrative or psychological load. The comparison with near-miss rates and control verification helps prevent a new digital metric from becoming another isolated lagging measure.

Technology-first rollout versus signal-first governance

The difference between a technology-first rollout and signal-first governance is visible in the questions asked before deployment. The first approach starts with features and adoption targets, while the second starts with the decision that must become faster, clearer, or more reliable.

ISO 45001:2018 places operational planning and control inside the occupational health and safety management system. ISO's standard overview supports a useful interpretation here. A technology pilot is not separate from the management system because its outputs can influence planning, operational control, competence, and evaluation.

The comparison also exposes a common trap. A high adoption rate can be reported as success even when the worksite has not changed. The stronger question is whether the people closest to the hazard can describe what they do differently because the signal exists.

Where human judgment must remain

Human judgment must remain in safety decisions that depend on context, timing, competing hazards, or the authority to stop work. Cam Stevens said that technology can elevate the human experience, while also recognizing that there are times when technology should not be used.

That boundary is important because a model may be trained on yesterday's conditions while a crew is facing a new task, a degraded control, or an unexpected interaction between hazards. Cam Stevens put the boundary plainly on the podcast: “We can elevate the human experience with technology, but there are certainly times when technology should absolutely not be used.” The signal can inform the decision, but it cannot assume the authority that belongs to the person who understands the work as it is about to happen.

Andreza Araújo's book The Illusion of Compliance provides a strong lens for this problem. A process can look complete because every field is populated, yet still fail to protect people when the operating decision is delegated to a record that no one challenges.

Leaders should define at least 2 non-negotiable human checks before deployment. The first should confirm that the person receiving the signal can question its meaning. The second should confirm that the person can stop or change the work without waiting for a system status to turn red.

What a 30-day review should measure

A 30-day technology review should measure decision quality, response latency, field change, and unintended work burden instead of counting activity alone. The review window is long enough to observe several operating cycles while remaining short enough to pause a weak pilot before it becomes infrastructure.

Cam Stevens's problem-first position gives the review a clear sequence. During days 1 to 7, define the exposure, the decision owner, the excluded uses, and the evidence required. During days 8 to 21, observe whether the signal changes supervision, escalation, or control verification. During days 22 to 30, decide whether to scale, redesign, or stop.

A useful review can ask 5 questions. Did the signal reach the right role? Was the response made within the agreed window? Did the field condition change? Did reporting become easier or harder? Did the pilot introduce workload, surveillance, or communication problems that were absent from the original approval?

The review should also preserve a record of false positives and missed signals. That evidence prevents a sponsor from presenting only the successful alerts and gives workers a credible route to challenge the system when it conflicts with what they see.

Recommendation

Use Episode 15 as a governance prompt for one live safety technology pilot, not as a reason to purchase another tool. Select 3 initiatives that already produce alerts, then ask each sponsor to write a one-page decision brief with the problem, accountable role, 24-hour response clock, human override, and field verification method.

After 30 days, keep only the pilots whose signals changed a decision or reduced uncertainty in the work. Pause any pilot that produces more activity without stronger control, because scaling an unclear signal only spreads unclear accountability across more people.

This recommendation aligns with the way Andreza Araújo connects engineering, creativity, and care. A technology decision is sound when it protects the person doing the work, improves the manager's judgment, and creates evidence that can withstand operational pressure.

Conclusion

Cam Stevens's contribution to Episode 15 is a governance test for safety technology. A signal deserves investment only when leaders can state the problem, name the owner, define the response clock, and prove that the work changed.

More data does not automatically create more control. The safer decision may be to redesign the workflow, limit the tool, or decline deployment when technology would weaken human judgment and speaking up.

Listen to the full conversation on the Headline Podcast with Cam Stevens.