Quick note from me
AI hasn’t replaced UX research, but it has changed how researchers work. The strongest research portfolios now show how you think, and how you work when AI is part of the process.
Today, I’ve invited Aadil, a design researcher from IBM who will break down 5 tips for crafting UX research portfolios in the era of AI. Here we go.
Story outline
Tip 01: Show impact, not activity
Tip 02: Turn signals into decisions
Tip 03: Make your synthesis traceable (Human > AI)
Tip 04: Tell the story in a way that gets shipped
Tip 05: Show a research library your team actually used
Guest spotlight
Mobbin’s MCP
Tip 01
Show impact, not activity
Many research portfolios show activity but not impact. In an AI-era workflow, you can stand out by showing how AI-supported exploration led to better decisions and how you managed risk.
AI can help you scan a domain, spot patterns, and generate early hypotheses. Make clear what decisions it unlocked, what could have gone wrong, and what improved.
What to include in your research portfolio:
The decision at stake
How AI accelerated exploration (prompts, sources, scope)
The risks you mitigated (bias, hallucinations, missed segments)
What you validated with human judgment and real evidence
The outcome (what changed, what you avoided, what improved)
Example
Tip 02
Turn signals into decisions
AI can flag issues fast, but your value is turning signals into the right decisions.
Lead with the decision + what changed
What was the product decision at stake?
What changed in the design, roadmap, or priorities because of your evaluation?
What did you prevent (support tickets, drop-off, rework, accessibility risk)?
Show your “human in the loop” method
What AI did (heuristic scan, UI pattern detection, copy checks, accessibility suggestions)
How you validated it (user sessions, analytics, benchmarks, expert review)
How you judged severity and impact (who is affected, how often, how costly)
Artefacts recruiters want to see
Before/after screenshots with annotations
A prioritised issues table with rationale (impact × effort)
A short “how I used AI” box: tool, prompts, guardrails, and where you did not trust the output
Example
Tip 03
Make your synthesis traceable (Human > AI)
Great research isn’t the insight. It’s the traceable path from messy evidence to a decision the team can trust.
AI can speed up clustering and summarising, but recruiters want to see your judgment:
How you turn raw notes into themes and priorities
How you separate signal from noise
How you reduce bias before recommendations land
What to show (credibility proof)
Inputs: sessions, data types, and questions
Your synthesis rules: tagging, clustering, merge and split decisions
Where AI helped: summaries, early clusters, quote retrieval
Where you overrode AI: what you rejected and why
Checks: contradictions, outliers, segment differences
Decision link: top insights and what changed
Example
Tip 04
Tell the story in a way that gets shipped
A research portfolio is not a report archive. It’s proof your work changed a decision and moved outcomes.
Lead with the decision and the outcome
What decision was blocked, risky, or unclear?
What changed because of your work (priority, scope, flow, messaging)?
What improved (activation, conversion, time-on-task, errors, satisfaction, support tickets)?
Tailor the story to the people who had to act
Who was the audience (PM, Design, Eng, Execs)?
What did they care about (risk, speed, revenue, feasibility)?
What format worked best (1-pager, readout, storyboard, PRD inputs), and why?
Use AI to speed up packaging, not replace judgment
What AI did (outline, first draft, slide draft)
Your guardrails (sources, verification, rewrites)
The “so what” you owned (recommendation, trade-offs, next steps)
Example
Tip 05
Show a research library your team actually used
A good research library is a simple system that helps the team reuse insights, avoid repeat work, and make better decisions.
Start with the value
What did it help the team do faster or better?
Who used it (Product, Design, Engineering, Support, Leadership)?
What changed because of it (roadmap, priorities, fewer escalations)?
Show how it works
Tags: how you label by theme, segment, journey stage, and confidence
Findability: how someone gets an answer in under 2 minutes (search, filters, templates)
Reuse: how new studies link back to past evidence and decisions
Ownership: who maintains it and what gets archived
Explain how you used AI
What AI did: suggested tags, grouped themes, summarized studies, improved search
What you checked: sources, confidence labels, review steps, where you changed AI output
What to include
A screenshot of the structure (tags, views, templates)
A 5-step “how to use this” guide
One example where reuse led to a shipped outcome
Example
⭐️ Guest spotlight
Aadil is a Design Researcher at IBM
And co-founder of Design XP, where he runs UX apprenticeships and foundational design/research courses. He also teaches UX at Toronto Metropolitan University and spends a lot of my time mentoring designers. You can find Aadil sharing more about UX research, portfolios, and career growth on Instagram and Linkedin.
🫶 Together with Mobbin
The one thing I do before generating anything with AI
I pull references first, before I write any prompt or generate any visual.
Mobbin has just launched an MCP server, and now Claude pulls real shipped screens from their library of 621,500+ apps directly inside my chat.
I describe what I’m designing, Claude pulls 5-10 references from apps like Stripe, Linear, Duolingo, then designs from those. Works with Claude, Cursor, Codex, v0 and other. Setup takes under a minute, available on Pro plans.
What I use Mobbin MCP for:
Researching how shipped apps handle a specific pattern (paywalls, onboarding, empty states)
Building moodboards from a client brief quickly
Critiquing my own screens against best-in-class apps in the same category
Grounding every AI prompt in real references before generating anything
Backing design decisions in PRDs and specs with real-world precedent
No more tab switching, searching, or screenshotting one reference at a time.
And to be fair, curating inspiration and researching patterns has been my favourite part of designing with AI. And the most transformative.
My Designer Toolkit 🛠️
Tools I actually use in my workflow. Some links support this newsletter.
Want help with your UX portfolio? 🎁
Build your UX Portfolio with this course
Book a portfolio strategy call with me
Questions? Reply directly.
Keep designing ✨
Aneta
