Asset Modelling 3.0: Making complex systems legible
Modern asset systems are powerful, yet nearly impossible to read. This is the story of turning a database viewer into an interactive system map so teams can see cause and effect instead of memorizing it.Modern asset systems are powerful and nearly impossible to read. So I turned a sleepy database viewer into a living system map, where teams can see cause and effect instead of memorizing it.
Systems built for power, not for people
Enterprise asset platforms hold the data that keeps critical infrastructure running, yet the people who depend on them could barely navigate them. Context was scattered across spreadsheets, tickets, and a couple of senior engineers' memories. Assets connected to other assets, but those connections were invisible by default, so when something broke, tracing the cause took hours.
People don't struggle with data. They struggle with relationships.
The old interface answered what: rows, columns, IDs, status flags. It never answered why. What users actually needed was cause and effect: a view of how one asset affects everything downstream. That meant a different mental model entirely, moving from "a list of things that exist" to "a network of things that interact." The design response was to make that network visible, explorable, and trustworthy.
From a database viewer to a system map
A relationship workspace
The canvas shows the whole asset network at once. Click any asset to see what's at risk downstream, understand impact before making a change, and read system health from colour-coded states, no query required.
Guided modelling
Building a model used to demand deep system knowledge. A scoped, step-by-step flow replaced that with structure, so inline validation catches mistakes before they propagate and new users model independently in their first session.
AI as operator
Ask "why is Tower A failing?" and the graph focuses the asset, surfaces the upstream dependency chain in sequence, and suggests one-click fixes. AI that acts on the model, not a chat log beside it.
Trust as a design decision
Every element reports its status; uncertainty is never silent. Errors surface inline where users can fix them, and a completeness score tells people when a model is actually ready to be trusted.
What changed
The system didn't just improve the tool: it redistributed knowledge across the team, so expertise stopped living in two or three heads.
Complex systems don't need more data. They need clarity, and clarity is a design decision.