Defense-adjacent geospatial intelligence
A defense-adjacent geospatial intelligence team brought Bayesian Leaf in to fix onboarding for one product. They kept expanding the mandate, from that scoped project to a standing fractional Head of Design role.
Bayesian Leaf is the practice of Hew Suber.
The opaque system
The client's analyst platform surfaced answers but hid its own reasoning. Someone new landed in a workspace with four competing workflow entry points on the welcome screen, data-source toggles and engineering debug views exposed in the interface, a tab bar that buried the answer, and a "search" that was really a chat box with no visible scope.
You could get an answer. You couldn't see how the system reached it, or steer it. The decisions were invisible.
The read
The instinct is to treat this as a design cleanup. It wasn't. Every complaint traced back to one thing: the product made decisions the person couldn't see. The work was to make those decisions legible.
Made legible
The direction surfaced what the product was doing. Scope shown once, in dedicated widgets for location, time, and topic, instead of repeated everywhere. The reasoning moved into the workbench as one line that expands into the full step list. The answer became the default landing state. A guided onboarding layer taught the product's own anatomy: how it reasons, how to refine scope, where the evidence lives, how the map works. And a monitoring model let a person watch a search over time instead of re-running it by hand.
✿Confidence indicators and semantic grouping were part of the same move: making the system's judgment visible, not decorative.
Shipped and adopted
The prototype became the spec the engineering team rebuilt the product from. The welcome-screen cleanup shipped.
And the altitude was set early. Less than five weeks into the engagement, Bayesian Leaf was presenting design direction directly to the CTO and CEO. The monitoring direction shown in that early presentation was built months later. Set the direction, watched it ship.
The moment it inverted
The CEO hit a wall of individual UX problems using the product on a live task. Instead of triaging them as a bug list, Bayesian Leaf collapsed the whole pile into one diagnosis.
One reframe turned a mess of complaints into a single, addressable direction. That is decision infrastructure made visible, in the room, in front of the person who runs the company.
Outcome
The direction shipped and became how the product works.
A scoped onboarding project became a standing fractional Head of Design role.
The CEO began dogfooding the product on real work and bringing UX problems straight to Bayesian Leaf, which was pulled into the recurring product sync. The relationship inverted: from "here's a project" to "help us decide where the product goes."
The numbers aren't here yet. Post-change usage is still being measured, and when it is, it goes here.
Bayesian Leaf didn't redesign screens. It took a system whose decisions were invisible and made them legible, so the team could see what the product was doing and steer it. That's the work: not design, decision infrastructure.
The one question
Can the people using your product see the decisions it's making for them?
If the answer is no, that's rarely a design problem.
Let's talk