Intelligence Hub
I turned a workflow spread across email, phone calls and spreadsheets into one shared case workspace for six intelligence analysts.
Internal case-management work during my public-safety internship. The hard part wasn't the screens. It was six stakeholders, sensitive workflows, and a product that had to remember everything that happened before.
Role
UX design intern
Stakeholders
Six intelligence analysts
Timeline
Summer 2026
Status
In use by the team
Screen: the final case workspace, sanitized
Quick facts
I owned: stakeholder interviews, workflow mapping, product structure, interaction design, usability testing.
Constraint: sensitive workflows, so everything shown here is sanitized.
Built first in Google Apps Script so the team could try it safely. The team has since started its own server version.
The problem · Decision 01 · Decision 02 · Decision 03 · Outcome
The problem
A case's history lived in three places at once.
Information came in by phone and email, then got tracked in spreadsheets. It looked like an intake problem. It was really a continuity problem: there was no easy way to see the full history of a case or a person without rebuilding it from different places.
Screen: before map, email + calls + spreadsheets (sanitized)
Decision 01
Make history a first-class part of the product.
The easy option was a nicer tracker. I pushed for something different: treat cases, people, ownership, status and past activity as connected parts of one workspace. That changed the structure of everything that came after, because any screen could now answer "what happened before this?"
Screen: relationship model, person → case → activity
Decision 02
Treat disagreement as product data.
Six analysts weren't six copies of the same interview. I watched who answered first, who hesitated and who stayed quiet, and when one voice took over I changed the question or brought someone else in. The product depended on hearing the whole team's workflow, not just the loudest version of it.
Screen: two analysts, two behaviors, and how the design handled it
Decision 03
Optimize for context recovery, not feature count.
As the surface grew, I compared scrolling, tabs, top-level navigation and different hierarchies. The test wasn't which looked cleaner. It was how fast an analyst could tell where they are, what changed and what to do next. That made me much stricter about navigation, visible status, ownership and how much to show at once.
Screen: before
Screen: after
Iteration + AI
AI helped with: implementation directions, variants and moving through alternatives fast.
I decided: the product model, scope, hierarchy, and what stayed. I kept the model stable and was very willing to throw away the screen built on top of it.
Outcome
From scattered tracking to one workspace.
The concept brought history, status, ownership and related information together. The first version ran in Apps Script so the team could test it safely, it's in use by the team, and they've started building their own server version. They still use separate monitoring apps alongside it.
Screen: final flow (sanitized)