◆ My Role · Product Designer
I led research synthesis, translated field insights into tablet and desktop workflows, built the design system, and shipped an interactive prototype with Claude Code; and then tested and iterated it directly with investigators.
ORGANISATION
Evidence Linking
Primary and secondary relationships
DATA ENTRY
Guided Capture
Required fields and AI prompts
RECORDS
Chain of Custody
Starts at the scene
SECURITY
Verified Handoff
Face ID before print and submit
TEAM
ROLE
TYPE
TIMELINE
SKILLS
01 · THE PROBLEM
Crime scenes are
chaotic
, but evidence records need
to be precise as one missing detail can cost
justice
Documented under pressure
Officers record evidence in stressful, time-sensitive scenes where attention is scarce.
Built on paper
The process still runs on handwritten forms, manual entry, and disconnected systems.
Room for error
That opens the door to missing details, illegible notes, duplicate work, and chain-of-custody gaps.
◆ WHY IT MATTERS
In the O.J. Simpson trial, questions about how evidence was handled and documented gave the defense room to build reasonable doubt. It's a lasting reminder that cases are won or lost on records — not just on what gets collected.
02 · ABOUT THE CLIENT
SampleServe had the foundation in
environemental evidence collection
SampleServe had already built tools for field-based sample collection. We explored how that same foundation could evolve into a crime scene workflow, helping officers document, label, link, and hand off evidence more reliably.
Before designing, we had to understand the
real proccess
03.1 · COMPETITIVE ANALYSIS
Most tools managed evidence
after
it entered
the system
We analyzed Axon, Omnigo, and NICE to understand how existing platforms support evidence management. This helped us identify where tools were already strong, and where scene-side evidence capture was still underserved.
◆ RESEARCH INSIGHT
Competitive analysis showed that most tools manage evidence after collection, with limited focus on the collection process or ensuring evidence is captured correctly at the scene.
03.2 · ARTIFACT STUDY
Static forms
couldn't adapt
to different evidence types.
We studied existing evidence forms to understand what information officers were expected to capture across different evidence types. This helped us define design requirements for a dynamic form system, where repeated fields could be auto-filled and evidence-specific fields could appear only when relevant.
◆ RESEARCH INSIGHT
A one-size form forces officers to wade through irrelevant fields.
The fix: a dynamic form that adapts to the evidence type and auto-fills what's already known.
03.3 · INTERVIEWS
Interviewed 6 officers to
understand evidence
collection from field to the courtroom
We interviewed 6 law enforcement professionals to understand how evidence is collected, documented, handed off, and later defended in real workflows. The goal was to learn where officers lose time, rely on memory, or work around the current process.
◆ COLLABORATION
I partnered directly with investigators and forensic stakeholders — interviewing them, mapping their handoffs, and validating language. Their input shaped the workflows and the terminology the product uses today.
Once we finalized the user flow, we designed and
built v1 prototype using AI assisted workflows
06 · USABILITY TESTING
Observed what users did, what they said, and
where the workflow slowed down.
We looked at the test from multiple angles, not just whether users completed the task. We tracked repeated behaviors, captured participant quotes, and measured flow-level friction. Together, this helped us understand both the visible and invisible parts of the experience.
08 · FINAL PRODUCT
The final product turns a fragmented evidence
process into
one connected workflow
09 · DESIGN DECISIONS
Key choices that shaped the final experience
01
Rather than one long form, fields appear based on evidence type. Tier 1 fields (always required) auto-fill from the case; Tier 2/3 fields show only when relevant — reducing cognitive load at the scene.
02
Testing showed "Parent/Child" confused officers. Renaming to "Primary/Secondary" matched their mental model and removed friction from the most-used linking interaction.
03
Rather than a static report, the chain of custody updates in real time as evidence is collected, transferred, and handed off — making it prosecutor-ready by default, not as an afterthought.
10 · LEARNING
My biggest learning was knowing
where AI
helps,
and where
human judgement matters
AI was most helpful when the task was repetitive, technical, or execution-heavy. We used it to create documents, support scripting, and build interactive HTML prototypes, which helped us move faster and test ideas more tangibly.
But we were careful not to use AI as the decision-maker. Research synthesis, usability insights, and product direction still needed researcher judgment, team discussion, and context from the field.
This taught me to use AI mindfully: not to replace thinking, but to reduce busywork and create more space for better design decisions.
