Dots app shown across four screens: the launch dial, onboarding, voice capture, and a trivia recall moment

Dots Social Memory Aid

Concept Case Study · Mobile App Design

Role

Product Designer & Brand Strategist (End-to-End)

Time

2-Day Sprint

Type

Mobile Application

Dots is a social memory aid for anyone from a nine-year-old learning their cousin's world to a grandparent keeping up with a family that keeps growing. Capture a person in seconds after you meet them. Get them back the moment you need them.

Scope: Product Strategy, UX Architecture, UI/Visual Design, Brand Identity, Cognitive Interaction Design

Deliverables: Concept Definition, Ecosystem Architecture, Dynamic Color System, Wireframes, High-Fidelity Interactive Mockups

Dots launch screen on a single phone mockup

"We meet dozens of important people every week. Our brains aren't built to hold every name, story, and detail: forgetting doesn't mean you don't care, it means you're human."

Dots bridges the gap between caring about someone and actually remembering their story.

Problem

Forgetting isn't a character flaw: it's a memory-architecture problem. The same friction repeats across every part of life: a small, human detail slips before it's needed again.

Capture is too slow

By the time you'd open a contact card and pick fields, the moment, and the detail, is gone.

Recall is the wrong shape

Contacts are indexed by name. Humans index by story: the BBQ, the trade, the favor offered.

The cost is social

Forgetting isn't neutral. It reads as "you don't matter": to a coworker, a niece, a neighbor.

Example capture cards: Devon the doorman, Marco at the coffee shop, and a cousin's toddler

Insight

Color categorizes faster than reading does.

Dual-coding theory says a visual channel and a verbal channel together beat either alone. So every dot in Dots carries both: a color token your visual cortex sorts in milliseconds, and a plain-language line you read only if you need it.

The practical payoff: you pre-filter a network of 300 people before you've read a single name.

Family

Parents, siblings, cousins and everyone attached to them, kids and pets included.

Work

Coworkers, leadership, the front desk, and the conference contact you'll email next week.

Friends

School, childhood, and the ever-expanding ring of friends-of-friends.

Community

Neighbours, the barista, the dog park, the run club. The people you see weekly.

Partner

Your partner's side: in-laws, their siblings, and the names you're expected to know.

How Dots Works

01 · Capture

Say it, or type it

Ten seconds of speech in the driveway, or the same parser from the keyboard, for a quiet room.

02 · Map

See the whole web

A radial orbit: you at the centre, one hub per colour, every dot hung off whoever introduced them.

03 · Ask

Search by the story

"The guy at Alex's BBQ who does landscaping" is a valid query.

04 · Arrive

Walk in prepared

Before an event, Dots surfaces who'll be there and what you last talked about.

Inclusive Interface

Toggle these: the card on the right responds.

Wireframes

Every screen from the sprint, in sequence.

Launch: the whole idea in one frame.

Onboarding: one dot, then you're in.

Capture by voice: say it, done.

Capture by typing: the quiet option.

Map: your whole world, pre-filtered.

Ask Dots: search the way you remember.

Profile: the details, not the data.

Next Steps

What this 48-hour sprint didn't have time to prove, and what would need to be true before it ships.

Immediate

Stress-test at scale

Does the radial map still read at real family scale?

Every screen so far shows 3–5 dots. The open question is whether it holds at 300+ dots across four generations, or needs clustering or a search-first fallback once a category gets crowded. Untested: the single biggest assumption in the concept.

Validate voice-capture accuracy

Can "speak plainly" actually be parsed reliably?

The product hinges on turning a spoken description into name, category, and detail. Next step: a Wizard-of-Oz test: record 10–15 people describing someone they just met, and manually check how reliably that split comes out. If accuracy is shaky, capture needs a confirm-before-save step not yet designed.

Near-term

Design the empty state

How does the value land before a single dot exists?

Every screen here assumes data already exists. The real first-run experience (zero dots, zero contacts) is where the value proposition either lands or doesn't.

Design for work, explicitly

Does the same color-and-story model hold up under different social stakes?

The sprint scoped workplace relationships as one category of five. A standalone flow for professional networking would test whether the same model holds up directly, rather than staying implied.

Make privacy the headline, not the footnote

Would someone trust this with a boss's or in-law's details?

"Nothing left your phone" and "answered on device" are strong signals currently buried in small captions. People will hesitate to voice-log details about someone they don't fully trust the app with. That needs to be stated up front, not implied.

Make "details, not data" a stated promise

What's actually stored, where, and how would someone find out?

Right now it's a tone, not a policy. Before this ships, it needs a plain-language answer: no selling, no ad targeting, no training external models on someone's family details, deletion that's actually instant and complete. A broken trust claim is worse than no claim at all.

Next Project