←Work/Reimagining identity profiles with AI-driven insights

Reimagining identity profileswith AI-driven insights.

An identity could appear a half-dozen times across integrations. I unified the picture and folded in Gen-AI summaries without making people read a wall of text.

Project metadata

Client: Red Canary
Role: Lead Product Designer
Team: 1 PM · 4 Eng · 1 Designer
Shipped: Q3 2025
Type: StrategyRefactor
Result: -15% time-to-decision

Friction

— Pain points for the customer

↗Customers said identity pages were unhelpful and were comparing them unfavorably to competitors.
↗Important identity context was buried in long pages instead of surfaced as actionable insight.

My Role

— What I owned during the process

↗Owned the full design process from research and ideation through prototyping and handoff.
↗Pushed for forward-looking AI features without losing the core investigation workflow.

Outcome

— What I shipped

↗Delivered parity with competitors while introducing new AI-driven insights.
↗Cut analyst time spent correlating threats by roughly 15%.
↗Created a summary pattern that leadership and customers both responded well to.

The TL;DR.

Summary

I led the redesign of Red Canary's Identity Profiles, introducing actionable insights powered by generative AI and a modernized user experience. The update strengthened customer retention and gave us a significant competitive edge in a previously stagnant area of UX.

Flat. Boring. Behind the curve.

The Problem

The original Identity Profiles were a wall of raw activity. Customers found them unhelpful, and compared with competitors' identity tools, the experience felt behind in both utility and perception.

The team also had strong in-house GenAI capabilities, but the real design problem was deciding how to use them without making analysts read a novel in the middle of an investigation.

Key Decision

"How much GenAI content is too much?"

We had an in-house GenAI product with a lot of untapped potential. It could generate plenty of content, but that did not mean customers needed to read all of it.

The Decision

Don't make people read (oh the irony).

Use GenAI to create brief, high-value summaries only, while keeping the underlying evidence available when customers want to dig deeper.

Click to unlock →

Less reading. More answers.

The Solution

The redesign focused on surfacing actionable insights instead of making people parse raw data first.

Benchmark the category

I reverse-engineered the identity flows from the teams we needed to beat, then separated the expected baseline from the places we could move the experience forward.

Use AI where it helps

GenAI summarized and contextualized identity activity directly in the product, but only in short, high-value moments that accelerated decisions.

Keep the evidence visible

The old information still had to be available somewhere, so the page balanced quick insight up top with deeper supporting detail below.

Five months. Six iterations.

The Process

Exploration

Low-fidelity wireframes mapped the core layout and key interactions before any visual decisions were locked in.

Early structure

Wireframes focused on the two-column behavioral split (consistent signals left, anomalies right) before any visual polish was added.

Step 01

Competitive benchmarking

I started by mapping the products customers were already comparing us to so we knew exactly where the floor was and where we still had room to differentiate.

Step 02

Designing and prototyping

Low-fidelity prototypes helped validate direction early, then high-fidelity work tightened the interaction model and the right amount of AI assistance.

Step 03

Customer-driven validation

Pilot feedback shaped the final IA and a tiered rollout let us capture usability issues before the broader launch.

Shipped

The final experience, tested with customers and validated through iteration.

Benchmarking the field

I mapped every competitor's identity experience to establish the floor (what customers already expected) before deciding where to move ahead of it.

AI summaries, not essays

GenAI generates a short, high-value wrap-up of identity activity. Enough to act on. Not enough to drown in.

Evidence stays visible

The raw data didn't disappear; it moved below the fold so analysts can still dig in when they need to.

The shipped experience

A unified identity profile that reduced analyst correlation time by roughly 15% in early customer testing.

What worked. What got tricky.

Reflection

Tough spots

We iterated heavily on how much LLM content was useful versus distracting.
Existing information still needed a home, which made the page hierarchy trickier than it looked.
Customer expectations were already shaped by competitors, so the bar for usefulness was high from day one.

What went right

Having a wealth of usable data made the AI layer genuinely valuable instead of decorative.
The identity wrap-up summaries landed well with both leadership and customers.
Phased release planning helped the team get the MVP out sooner and learn continuously.

< Previous

Personal Project - 2024