The AI Market Has a Clarity Problem
I've been writing about something that we've been watching for years. Industrial enterprises have invested heavily in digitizing operational data through equipment sensors, core enterprise and industrial systems, and a broad range of data and BI platforms. Yet, despite these huge investments, in most operations the important decisions are still made outside those systems.
Attending HumanX 2026 this past week provided a useful moment to step back and assess where the AI market is today in this context, not in theory, but in practice.
I'll save the story of the event's AI-enabled conference "dating" program taking a copy of my resume and SteelTree description and somehow concluding that I spent 30 years at Rolls-Royce and now run a small telecom advisory firm called SteelTree in Swarthmore, Pennsylvania (specializing in cell towers across the U.S., Caribbean, and Latin America) for another time. (Apologies to Rolls-Royce Peter, and Swarthmore SteelTree too!)
What I came away with is that the AI market is not lacking capability, it's lacking clarity.
AI has entered what feels like a peak noise phase. Peak might even be premature, who knows, but it certainly feels like every company is now "AI-native," "transformational," and "redefining the future." When you look more closely, messaging is repetitive, differentiation is thin, and signal is hard to find. There is an enormous amount of activity, but much less clarity on what actually matters.
Despite the noise, some things are becoming clearer. There is much less discussion about AGI, model performance, and theoretical capability, and more focus on real-world deployments, ROI, and operational impact. That is a positive shift. AI is moving from possibility to accountability, with meaningful growth in real usage, not pilots or evaluations, and significant commitment to further adoption and investment.
So where is the gap?
Most industrial companies have their data digitized. Insight is increasingly available, and AI is more powerful than ever. But they are still largely left with the question of what to do next, and that is where the real gap exists.
Company systems today fall into two categories. Systems of record capture what happened. Systems of insight explain what happened. Neither are built to participate in decisions as they happen. That matters not because it is a technology problem, but because it is an operational performance problem.
Every day, decisions are delayed, issues escalate unnecessarily, experienced workers fill gaps manually, and outcomes vary depending on who is on shift. The result is lost productivity, increased cost, inconsistent performance, and avoidable risk.
The opportunity that emerges from this reality is not in more data. It is in improving what happens after the data is seen. When teams can understand what is happening in real time, make faster and better decisions, coordinate action across people and systems, and capture what worked, the impact is immediate:
- Faster issue resolution
- Reduced downtime
- Improved safety
- More consistent execution
- Measurable operational gains
Closing the execution gap means fewer delays between signal and action, fewer missed or inconsistent decisions, and better outcomes from the same underlying operation.
When decision-making becomes part of the system
When this works, when decision-making becomes part of the system, teams spend less time figuring out what to do, decisions become more consistent across shifts and locations, expertise is captured and reused instead of lost, and performance improves without adding complexity. This is where the ROI comes from, not from more data, but from better, faster, more consistent action.
Bottom line, for industrial organizations, AI today is not short on intelligence. It is short on systems that help teams act in the moment. That's the gap, and for organizations focused on reliability, safety, and efficiency, that's where the value is.
That's exactly where SteelTree is focused.