Applied research into adaptive, generative operational interfaces: how a system should present a complex decision by composing the right view for each task, instead of forcing the work into a fixed dashboard.
For decades, software has shipped a fixed set of screens and asked people to bend their work to fit. When the system can compose the right view on demand, that constraint disappears, and adoption follows.
The first television programmes were radio shows with a camera pointed at them, because nobody could imagine the new medium yet. Most AI interfaces today are the same: old dashboards with a chat box bolted on. The interface should instead be built for the decision in front of you.
Fixed, prebuilt screens. Predictable and trusted, but they force every task into the same shape.
The system describes the view it needs and a trusted engine renders it. The pragmatic middle ground.
The system composes the interface for the task on demand, contained safely, with a person able to verify it.
The right information, at the right time, explained clearly enough to act on without second-guessing.
Project Aperture studies how an operational system should present itself: how much of the interface should be composed for each task, what stays fixed for trust, and how a person keeps their bearings when the screen is no longer the same every time.
How the system builds the right view for a tender, a deal or a portfolio, rather than one generic screen.
How much should flex per task, and how much should stay fixed so people trust it.
How sources, figures and confidence stay visible, so an adaptive view is still a checkable one.
How people keep their mental model and can repeat a process as the view adapts.
How generated interface runs in a safe sandbox, never with a free hand over the system.
How a person can always trace what the system shows back to its source.
The system composes the right view for the decision, rather than forcing the work into a generic screen.
The interface flexes on top of fixed, trusted foundations, not a different screen every time.
Sources, figures and confidence stay visible, so an adaptive view is still a view you can check.
People keep their mental model and can repeat a process, even as the view adapts.
Generated interface runs in a safe sandbox, never with a free hand over the system.
Whatever the system shows, a person can trace it back and confirm it.
Firms working with us through Aperture get a clear view of where fixed dashboards are slowing their teams down, and what an interface composed around the decision, rather than the other way round, would change.
The winning systems will pair adaptive views with stable, trusted foundations, so people gain flexibility without losing their bearings or the ability to verify a result. The interface becomes part of the reasoning, not a wrapper around it.
Project Aperture is open research into operational interfaces: how a system should present a complex decision. We publish the questions we are wrestling with, the patterns that recur across real deployments, and the principles that hold up under pressure. We do not publish client data, or the parts of the method still being worked out. It is research, not a product: the questions are the interesting part, and the answers are earned in the work.