Seeing vs Searching

The value of operational views over reactive searching.

Most operational software is designed around helping users find information.

• You search for an asset.

• You open a customer record.

• You check a dashboard.

• You look at an alarm.

• You open a report.

Most of the time, this works perfectly well. The problem comes when operational decisions are needed. These rarely depend on a single piece of information. Instead, they require context across multiple locations, assets, documents and potentially millions of data points.

That context helps us understand why something happened, identify patterns, build more accurate models and reveal relationships that would otherwise remain hidden.

Perhaps more importantly, context solves one of the biggest weaknesses of search. If you don't know something exists, you'll never think to search for it.

To analyse operational problems effectively, it helps to understand the context surrounding each part of the operation.

• Where is it located?

• Which building or rooms does it serve?

• What maintenance has already been carried out?

• Are there any compliance inspections due?

• What documentation is available?

• Has its energy consumption changed recently?

• Are similar assets experiencing the same issue?

• What other types of events have occurred for this asset?

None of these questions are particularly difficult. The challenge is that the answers often exist in different systems, owned by different teams and presented in different ways.

As a result, people spend more time searching for information than making decisions.

Dashboards have become the default way of presenting operational information, and they are incredibly valuable. However, every dashboard is designed to answer a question that somebody anticipated in advance, and real operations don’t always work like that.

Unexpected events don't follow predefined workflows. When something unusual happens, people quickly leave the dashboard and begin searching across multiple systems to rebuild the context they actually need.

To address this, I believe we should think less about dashboards and more about operational views.

An operational view isn't another dashboard. It's a connected, semantically rich perspective that brings together everything relevant to a particular task or decision.

It may include:

• The people.

• The assets.

• The locations.

• The documents.

• The events.

• The measurements.

• The relationships between them.

Unlike a report, an operational view is not fixed. It is generated dynamically from an operational knowledge graph for a specific question or purpose. The same underlying operational model can therefore produce thousands of different operational views without duplicating information.

Importantly, this doesn't require every operational system to be consolidated into a single platform.

For example, a CAD drawing can remain inside AutoCAD. The operational view only needs to know that the drawing exists, what it relates to, its drawing number, where it is stored and how it can be retrieved. If somebody needs the actual drawing, it can still be opened directly from AutoCAD.

The objective isn't to duplicate data. It's to make the existence, location and operational relevance of information visible.

Once created, operational views become reusable building blocks.

They can support dashboards, reporting and automation by providing exactly the context required for a particular operational task.

For example, you might create an operational view showing every refrigeration asset across multiple factories, together with their energy consumption, maintenance history, service documentation and recent faults. Rather than building another bespoke dashboard, you've created a reusable operational perspective that can support multiple business processes.

Where I believe this becomes particularly interesting is when combined with LLMs and AI agents.

Instead of asking an AI to search disconnected systems and infer relationships for itself, we can provide it with an operational view that already contains the relevant semantic context.

Imagine creating an operational view containing every non-compliant asset across an organisation. An AI agent can immediately begin analysing relationships between assets, locations, maintenance history, inspections, responsible teams and operational events.

It may discover that the majority of overdue inspections relate to work originally allocated to an engineer who was absent for two weeks, and that no workflow existed to automatically redistribute those tasks. Rather than simply identifying overdue assets, the AI has identified the underlying operational cause and highlighted an opportunity to improve the process itself.

For me, that's where the real value lies.

The future isn't simply about making information easier to search.

It's about making the operational context available so that both people and AI can understand, reason and make better decisions from the information that already exists.