When the decision is the intervention

I wrote recently about why most organisations do not have an AI problem. They have a decision problem. The volume and complexity of decisions facing modern enterprises has grown dramatically, and the cost of getting them wrong has never been higher. The infrastructure most organisations rely on to handle those decisions has not kept pace. Better operating models or more data, on their own, do not close that gap.

What closes it is building a system where the decision itself becomes the unit of work. Where intelligence is translated into action at the right moment, by the right person, with enough shared context to act confidently rather than reactively.

I have seen this argument validated across many sectors over the past decade. But the example I keep coming back to lately is one I find genuinely striking, not because it is technically unusual, but because of the problem it is solving.

A problem that compounds quietly

Hospital patient flow is one of those challenges that is easy to underestimate. From the outside, it looks like operational metrics. ED waiting time. Bed occupancy rate. But behind each number is a myriad of interdependencies: bed availability, discharge timing, staffing ratios, transport coordination, pharmacy sign-off. 

And behind every metric is a person. A patient who has been told they can go home today but is still waiting three hours later for a medication sign off. A family waiting anxiously in the corridor, who took the day off to bring them home. A nurse fielding three phone calls about getting this patient home, none of them clinical.

What makes this problem genuinely hard is not that it lacks data. Most modern hospital systems have significant volumes of it. The challenge is that traditional discharge coordination was designed for a world of lower volume and simpler handoffs: reactive, manual, silo’d across departments. As patient volumes grow and healthcare systems become more complex, that model does not scale. It just slows down. A problem that compounds quietly, until it reaches a breaking point.  Not a single failure, but dozens of small misalignments. Each manageable in isolation, but collectively overwhelming. This is a decision problem. The information required to act well exists somewhere in the system. The problem is that it arrives too late, reaches the wrong people, or sits unconnected from the moment when it could change something.

What TeleTracking and University of Louisville Health did differently

I want to be careful about how I describe what I find impressive here, because it is easy to reach for technological language when the more important point is organisational. I should note that Faculty and TeleTracking are partners in this work, and we have been embedded alongside TeleTracking from the start. Which is perhaps why the human and organisational dimensions of what happened feel more significant to me than the technical ones.

TeleTracking a healthcare technology company, worked alongside their partner, University of Louisville Health (UofL), to do something that I think reflects a genuinely mature understanding of how decision intelligence works in practice. They did not simply implement TeleTracking’s platform and measure what happened. They committed to the harder work: embedding alongside frontline teams, redesigning workflows to connect insights to the moments where commitments are actually made and creating the conditions for honest feedback and continuous improvement.

The results, measured against a contemporaneous control site, speak clearly. Within five months, UofL achieved a 22% reduction in overall patient length of stay, a 20% increase in morning discharges, a 46% increase in ED in-patient admissions processed, and discharge processing that was 15 minutes faster. These were sustained improvements, achieved while facing increased admissions volume. For frontline teams, the change was tangible. A nurse who started the shift knowing which patients to prioritise for discharge. A patient home by lunch, as planned. A bed freed for the patient still waiting in ED.

Numbers like that do not just come from deploying a tool. They come from redesigning how a system makes decisions.

Why this matters beyond the hospital

Gartner's analysis of where healthcare is heading is worth sitting with for a moment. By 2030, they project that 65% of hospitals using modern EHR systems will depend on intelligent hospital operation centres for real-time tracking of critical business and clinical workflows. The same research names TeleTracking as a representative vendor in this space. 

These are not distant predictions. The organisations that will be positioned to act on them are the ones building the decision infrastructure now, not the ones waiting until the technology matures further.

What the UofL example illustrates is that this infrastructure is not primarily about data warehouses or dashboards or models. It is about creating a shared, forward-looking view of priorities across teams that have historically operated independently rather than together. Physicians, nursing teams, centralised hospital services, and the transfer centre all working from the same picture. Bottlenecks identified before they materialise. Discharge strategies tested in simulation before they are deployed. That is not a technology story. It is a coordination story enabled by technology.

The quiet insight

There is a pattern I have noticed consistently across the sectors where decision intelligence has taken root most effectively. The organisations that benefit most are not always the ones with the most sophisticated models. They are the ones willing to treat the decision environment itself as the thing that needs to change.

What TeleTracking and UofL have demonstrated is that this is possible in one of the most constrained, high-stakes, human-centred environments imaginable. If the reasoning that makes Decision Intelligence powerful can operate at that level of complexity, with that level of accountability, in an environment where the stakes are genuinely human, then the argument for it in other sectors becomes harder to dismiss.

I find this sort of application genuinely encouraging. Not because it validates a category or a product, but because it points to something more important: that the gap between intelligence and action, which I believe is the central problem facing most complex organisations today, is a gap that can be closed. The tools are available. The methods are established. What remains is the will to treat the operating model itself as the design problem.

The organisations that do that work now will not look dramatically different in the short term. But in five years, when the decision load has grown further and the complexity has compounded, the distance between those who built this infrastructure and those who did not will be very difficult to close.

Read the full story

The detail behind what TeleTracking and UofL Health achieved together is worth more than a summary. The full case study covers the operational context, the implementation approach, and the validated results in depth. I would encourage anyone working in healthcare operations or thinking seriously about where Decision Intelligence fits in their organisation, to read it.

If the UofL results raise questions about what this approach could look like in your organisation, TeleTracking has the right people to talk to. For over 35 years, they’ve focused exclusively on healthcare operations and orchestration, in one of the most demanding environments imaginable, and the results reflect that.