Unifying filtering across a security platform
Four inconsistent filtering patterns, unified into one AI-assisted system that users learn once and use everywhere.
Role: Sole Product Designer | Timeline: 1 month | Tools: Amplitude, Figma, Claude, Dia, Cursor
*All visuals use generic, non-company data for confidentiality.
Overview
A cybersecurity (WAF/API) platform had grown module by module, and its filtering had grown the same way: four different patterns across six modules, each with its own terminology.
On the critical path to the platform's core value, finding and fixing security issues, users were stalling.
I led a data-informed redesign that unified filtering into one consistent, learnable system.
It tested well with roughly ten users, earned strong early client feedback, contributed to closing new deals, and is now rolling out platform-wide.
Challenge
Six modules, four filtering patterns, four vocabularies.
A user who learned to filter in one module had to relearn it in the next.
Amplitude confirmed the cost: people clicked the same filters over and over without reaching a finding, and the path to remediate an issue was full of dead ends.
Because filtering sits directly on the path to the platform's value, this friction was quietly throttling adoption, especially for less technical users.
Discovery & Research
The problem surfaced during a routine Design QA pass I led.
I pulled Amplitude data to find exactly where users got stuck, then interviewed power users to learn what they were trying to achieve, how they filtered today, and how they filter in other tools they trust.
I benchmarked filtering conventions across the market, then synthesized everything into a set of behavior options to test.
Process
I explored several directions,
pressure-tested them in a product review, and ran simplified A/B comparisons with power users.
I used Claude alongside Figma as a living documentation and exploration space, and used Dia to build both a stakeholder presentation and an interactive prototype of the chosen direction.
That let me validate the full flow with roughly ten users in rapid, AI-assisted usability testing, compressing explore, test, and decide into a single work cycle before handing engineering a tested solution.
Stakeholder presentations I used to align the team on the chosen direction.
The Solution
One filtering system, consistent across every module, with a single shared vocabulary, built to serve everyone from a first-time viewer to a daily power user.
Ask AI: a natural-language search ("describe what you want to find") so non-technical users can filter without learning any syntax.
Customized per client: clicking the search bar opens a dropdown of tailored search suggestions.
Simple / Advanced toggle: progressive disclosure, a clean set of controls for newcomers and full depth on demand for power users.
Grouped Filters: related filters organized and collapsible, keeping complex queries legible at a glance.
Save View: save a filter configuration and return to it, removing repetitive setup for recurring investigations.
Figma was structured for a clean, design-to-code handoff,
and engineering shipped the system within a single cycle.
Impact & What's Next
Validated with roughly ten users and met with strong early client feedback, the work contributed to closing new deals.
It is now implemented and rolling out across all clients, instrumented with session recordings against defined flows so we can measure the lift.
Filtering became the first proof point for a broader effort now consolidating all six modules into one coherent product.
What I Learned
The hard part of a security product isn't adding capability, it's making complexity legible to users who aren't all security experts. Designing one system for a spectrum of technical fluency, and using AI to shorten the loop from idea to tested prototype, is how a solo designer moves a product this complex in a single month.