Web AppAI-Enhanced
Signal
Real-time earnings calendar for investors and traders
{ CASE STUDY }
Bundance
A browser-based AI-assisted shopping app that helps users find the right product faster - through chat, image upload, or voice.
Web AppAI-Enhanced
{ GOAL }
Design a fast, low-friction shopping journey
01intent
02relevant results
03retailer click-out
Help users go from intent → relevant results → confident retailer clickout — without needing an account.
{ SUPPORTING GOALS }
Three things this needed to solve
01
Reduce effort at the start
Let users express what they want through chat, image, link, or voice — instead of forcing category browsing.
02
Build confidence on the product page
Help users decide with a clear offer comparison: price, discount, delivery, shipping, retailer — plus quick AI guidance.
03
Create a trustworthy handoff
Make the redirect step transparent (opens retailer in new tab) so users feel safe leaving and can easily return.
{ USER PERSONA }
Two core user types
To keep the UX grounded in real behavior, I defined two core user types that represent the main shopping mindsets this product needs to support.
Maya
The Fast Gift Finder
PROFILE
Age 28 · Marketing Manager · Sydney, Australia
CONTEXT
Maya often buys gifts close to the deadline. She wants a confident recommendation without opening lots of tabs. She answers quick clarifying questions and clicks out once it feels right.
PAIN POINTS
Too many irrelevant results, not knowing which specs matter, low trust in unknown sellers.
SUCCESS LOOKS LIKE
3–5 strong options quickly, clear "best fit" signals, one-click path to a trusted retailer.
"I want a good choice fast."
Robert
The Deal-Driven Comparer
PROFILE
Age 34 · IT Specialist · Madrid, Spain
CONTEXT
Robert compares offers across multiple retailers and cares about the real total cost — price, shipping, and delivery date. He refines searches aggressively before clicking out.
PAIN POINTS
Losing track of what he already checked, misleading discounts, shipping fees not visible early enough.
SUCCESS LOOKS LIKE
Best deal is obvious (price + shipping + delivery date), shortlist items, watchlist and history to continue later.
"I need proof and the best total deal."
{ FLOWS }
Two distinct flows, one coherent product
Bundance serves two types of users — guests with limited access and members with the full experience. The guest flow is deliberately lightweight: get to results fast, drive click-out, nudge toward account creation. The member flow adds persistence — saved searches, watchlist, notifications, and search history.
Guest Flow
Goal: get users to results fast and drive click-outs. Limited to a set number of AI searches before being prompted to create a free account.
- Open the app → Landing page with AI search bar
- Search submitted → Loading animation while AI thinks
- Results page → 4-column product grid + AI chat panel
- Refine via chat → AI suggests categories as checkboxes
- Product detail → retailer comparison table
- Redirect interstitial → opens retailer in new tab
- Search limit reached → Sign up modal with integrated Sign In
Member Flow
Goal: persistent, personalised shopping experience. Members get significantly more AI searches, saved history, watchlist tracking, and notifications.
- Dashboard → AI greeting + search bar + recently viewed
- Watchlist → saved products with On Sale / Out of Stock filters
- Search history → all past searches with result counts
- Notifications → price drops, deals, watchlist updates
- Pricing → Free vs Pro upgrade at $4.99/month
- Feedback → structured 4-state feedback flow
Guest Flow
The light experience — no account required
Focused, fast, friction-free. Built to prove value and convert guests into members.
{ The first screen }
Landing
Nothing competes for attention. No sidebar. No navigation. Just a clean page with one purpose — search. The search bar sits centre stage. Users can type naturally ("gift for my boyfriend, tech lover, around €150"), upload a photo of something they want to find, or speak by voice. Three input methods, zero friction. Below the bar, suggestion chips — Mobile phones, Best value, Wedding, Home office — give first-time users a starting point without overwhelming them. At the bottom, the retailer logos aren't decorative. They answer the question every new user silently asks: "Is this real? Where will my results actually come from?" Seeing names they already trust makes Bundance credible before a single word is typed.
{ Search experience }
Results, refined by AI
The results appear. Alongside them, the AI chat panel stays open — persistent, following the user through the entire session. The user can browse visually and keep the conversation going at the same time. The AI asks a follow-up question, suggests categories as checkboxes directly inside the chat bubble, or explains why a product was surfaced. The grid updates as the dialogue progresses. It feels less like a search engine. More like someone who knows the market is right there with you.
{ Loading }
Designed for the "wait"
Bundance pulls products from multiple retailers in real time. That takes a moment — and a blank screen would kill the experience. So instead of a spinner, the wait becomes part of the story. A loading animation plays: product silhouettes scroll across the screen, keeping the user engaged while the AI works. It signals that something real is happening — not a cached result, but a live search across multiple platforms. The wait feels intentional. Not a delay, a process.
{ PRODUCT PAGE }
Decision clarity. Retailer handoff.
The user found something interesting. Now they need to answer one question: is this the right choice, and where should I buy it? The product page is built around that moment. A clear product summary at the top — images, name, key specs. Below it, an offers table showing price, discount, delivery date, shipping cost, and retailer side by side. No switching tabs. No guessing. Everything needed to decide, in one place. Once the user picks a retailer, a redirect interstitial appears. A simple modal that says: you're leaving Bundance, but your search stays open. One click to confirm. The retailer opens in a new tab. The user leaves with confidence. Bundance stays ready for when they come back.
{ Guest - conversion }
The moment that matters
When a guest hits the search limit, a modal appears. Not a wall — an opportunity. By this point they've already experienced Bundance. The value is proven. The modal simply captures that momentum and makes signing up the obvious next step. One smart detail: new users see Sign Up. Returning guests who forgot to log in click "Already have an account?" the form switches to Sign In without leaving the modal.
Member Flow
The full experience — persistent and personalised
Saved searches, watchlist, notifications, history. Built to reward returning users.
{ Now it gets personal }
Home. The starting point.
fter signing up, new functionalities are unlocked. A full dashboard opens: sidebar navigation, personalised content, and a familiar face: the AI search bar, still centre stage. The greeting is personal: "Happy Thursday, Alma." The dashboard doesn't start with data, it starts with the action. Below the search bar, recently viewed products let the user pick up exactly where they left off. The sidebar opens with full labels by default. This is intentional - new members need to see and understand every feature they've just unlocked. Once familiar, the sidebar can collapse to icons only, giving the content the space it deserves.
{ WATCHLIST }
Your personal product inventory.
Every product a member saves lands here. Not just a bookmarks list — a living inventory that tracks what matters to them. Three filters keep it simple: All, On Sale, Out of Stock. These cover the two moments members come back for - finding a deal and knowing something is available again. Nothing more, nothing less. The experience stays consistent from the first search to the saved collection.
{ SEARCH HISTORY }
Never lose a great find.
Every search a member runs is saved automatically. The name, the date, how many results it returned, and the category, all visible at a glance. One small detail worth noting: the result count. Before clicking into a past search, the user already knows if it's worth revisiting. 124 results for "Birthday gift." 11 for "New microwave." That number alone shapes the decision. No dead ends either. Every row links straight back into the results. Pick up exactly where you stopped.
{ Feedback }
Designed to be heard.
User feedback is one of the most valuable things Bundance can collect, especially at this stage. But most feedback flows feel like a chore. A long form, a rating out of ten, a text field that stares back at you. Users close it without a second thought. The approach here was different. A smooth, lightweight flow designed to lower the barrier completely. Four states: a calm starting point, a topic selection, a typing moment, and a confirmation. Each step is small enough that the user barely notices they've completed the whole thing. The goal was simple. Make it easy enough that people actually do it, and specific enough that the feedback is genuinely useful.
{ WHAT'S NEXT }
The road ahead
Bundance is at the beginning. Everything is shaped by user experience and refined constantly as AI evolves. New capabilities, new behaviours, new expectations. The product adapts alongside them. What feels right today will be challenged tomorrow, and that is exactly how it should be. The goal was never a finished product. It was a foundation worth building on.
{ LEARNINGS }
What this project taught me
01
Two flows means two products in one
Every shared element had to work equally well in both contexts without modification. Consistency is not about repetition. It is about trust.
02
AI needs a personality, not just a function
Tone is half the product. The difference between a tool and an assistant is not capability. It is how it speaks.
03
Simplicity requires more decisions, not fewer
Every element removed is a deliberate choice. The restraint is what makes the product feel confident.
04
Friction designed well becomes motivation
The search limit is the best conversion moment in the product, because by the time a guest hits it, they already believe in Bundance.
