All work
U-M Division of Public Safety & Security

BRIEFS

One building question could mean ten different tools. I saw it happen mid-shift.

Timeline

Summer 2026

Status

Prototype and requirements handed off

01 / The starting point

I sat with dispatchers during their shifts. Building information lived in PDFs, Dropbox and other tools. One workstation had seven screens, and a single lookup touched more than ten tools. To check whether a building contact still worked at U-M, a dispatcher had to jump into MCommunity too.

02 / What we found

Some contacts and building details were out of date. A new search box wouldn’t solve that by itself.

The prototype.

Find any building: type a name, number or nickname.
Prototype + PRD

Keep the urgent details where you can see them.

I designed a building profile with a pinned strip for contacts, access and hazards, plus predictable sections for everything else.

Prototype and requirements handed off. Not tested with dispatchers yet.

Design decisions.

01 / On the dispatch floor

Seven screens. Still searching.

I watched dispatchers work with PDFs, Dropbox and other tools. Even confirming a building contact could mean opening another system.

02 / The harder problem

Faster access to stale information isn’t enough.

Some contacts and building details were out of date. A new search box wouldn’t solve that by itself.

03 / The building profile

Keep the urgent details where you can see them.

I designed a building profile with a pinned strip for contacts, access and hazards, plus predictable sections for everything else.

04 / Where it landed

Handing off the prototype and requirements.

I handed the prototype and requirements to DPSS. It hasn’t been tested with dispatchers, so I’m not claiming time saved.

01

Hazards pin to the top, before anything else

Urgent information stays in view, reducing the need to hold it in memory while searching. This is a design rationale, not a measured result.

Hazards pin to the top, before anything else.
02

Jump straight to the section you need

A visible section rail lets people recognize where to go and skip repeated scrolling.

Jump straight to the section you need.
03

Every building, the same four tabs

The same tabs in every building profile make the layout easier to learn and predict.

Every building, the same four tabs.
04

BRIEFS Bot: ask in plain words

Plain-language questions can reduce navigation steps. Source citations let the dispatcher check the answer.

BRIEFS Bot: ask in plain words. It answers only from the records and cites them.
05

Fixing the source: old records moved into one clean sheet

Bringing records into a consistent structure reduces the need to reconcile information across tools.

Fixing the source: old records moved into one clean sheet.
More process notes

I sat with dispatchers during their shifts. Building information lived in PDFs, Dropbox and other tools. One workstation had seven screens, and a single lookup touched more than ten tools. To check whether a building contact still worked at U-M, a dispatcher had to jump into MCommunity too.

There were no design files to start from. My field notes also turned up outdated contacts and building details, which meant a faster search on its own would just surface the wrong record sooner.

Pin the urgent details

We chose: A persistent strip with the details needed mid-call, visible from every section of the profile

Over: Giving every piece of building information equal weight

Mid-call, a dispatcher needs contacts, access and hazards while they move through the rest of the profile. The strip takes up screen space, and I decided the mid-call task was worth it.

Based on what I saw during contextual inquiry. Not tested with dispatchers yet.

Predictable sections instead of one long document

We chose: Tabbed sections with a stable home for each kind of information

Over: A single scrolling document per building, like the existing PDFs

A dispatcher shouldn't have to remember where a detail is buried in a long file.

I went through several design-direction rounds and fixed sticky-header, tab-navigation and icon-meaning problems in my own prototype along the way. Reviewed internally; no dispatcher test yet.

A record has to be current, not just findable

We chose: Include checking and maintaining records in the scope

Over: Only making the lookup faster

Finding an outdated contact faster doesn't help anyone on a call.

From my field notes. Who maintains the records is still unresolved.

Contextual inquiry with the DPSS Dispatch Group; internal reviews of my prototype iterations.

Not tested: Dispatcher use of the prototype

Not tested: The 30-second lookup goal

Not tested: Who maintains the records

Where it landed

Prototype and PRD handed off to DPSS.

It hasn't launched or been tested with dispatchers, and nobody owns record upkeep yet.

What I learned

I learned to check where the information comes from before redesigning how people find it.

On a real shift, can a dispatcher find a detail, check that it's current, and trust the structure?

Next case studyOpen Library

Thank you for scrolling.

How can I help you?

Sonnet 4.6Low

Thank you for scrolling.

How can I help you?

Sonnet 4.6Low

Thank you for scrolling.

How can I help you?

Sonnet 4.6Low