philo

product design
interaction design
ux design

chapter 00

curated through intention

A narrow craft, a corner of science, a curiosity you picked up last month. Feeds bury these under whatever’s popular, and without a routine they fade until you remember them months later.

Philo gives them a daily home: three reads each morning, from interests you rank yourself, then it’s done. It’s built to be beautiful and to reward the attention you bring to it, because curating what you care about should feel like a choice you made, not a scroll you fell into.

thesis

Focused content. Expressive interaction.

Skip the prototype

Philo onboarding, with Film, Art and Music selected. — 0.1 live prototype

chapter 01

premise

fork: an infinite feed or three reads each morning
chose: three, then done
cost: every engagement lever a feed has
gain: completion

A feed or three reads, then done. Three gives range without overwhelm, and it finishes in fifteen to thirty minutes.

The edition for 8 October at rest: Film, Art and Music, three reads in 22 minutes. — 1.1home at rest, one slot from backlog

The Film slot opened: How Tati Rewired Cinema’s Relationship With Sound, marked as coming from your backlog. — 1.1home at rest, one slot from backlog

chapter 02

curation

fork: one fixed color per domain or accents that follow your picks
chose: the first three picks claim the accents
cost: no color consistency between users, and picks four to six stay neutral
gain: the interface shows your
hierarchy, not ours

Onboarding with five domains selected: Film, Art and Music lead in green, blue and pink; Design and Science broaden in black. — 2.1onboarding picks beside the home they produce

The edition those picks produce: Film in green, Art in blue and Music in pink. — 2.1onboarding picks beside the home they produce

2.2cut · the twelve-color palette

proposed, not built

Film#8BBF00
Visual Arts#0033EE
Music#CC0088
Architecture#E07020
Literature#7B2D8E
Science#0099AA
Technology#3355DD
Philosophy#8C6E3A
History#A03020
Economics#2D7D4F
Health#D4653E
Nature#4A8C2A

Twelve fixed colors collided at tag size. Your first three picks claim green, blue and pink, and the rest stay neutral.

chapter 03

Forty images, one publication.

fork: generate every image,
or source the real subjects
chose: sourced photos for real subjects,
generated images for concept pieces
cost: sourcing, licensing,
and grading every image
gain: the edition reads as one printed object

Sourced photos and generated renders, pulled into one edition by a shared grade and per-image fixes, so a morning reads as one issue, not forty sources.

The Art lane open: Vermeer and the Camera Obscura in full color, with Film and Music held above and below. — 3.1At rest, a duotone in the lane color. Open, full color in 0.58s. The Vermeer carries its own brightness after one global grade washed it gray.

chapter 04

saved and history

fork: one global press state or one per surface
chose: one per surface
cost: no single press language
gain: touch is where the delight goes

History cards bloom with light. Saved panels sink like keys. Home has no press state; the panel opening is the answer. Touch is the one place the restraint relaxes, and interaction has the impact to be truly rewarding.

History for October 2026: a pressed card blooms with light in its lane color. — 4.1History bloom, Saved keycap, home panel sequence

Saved: the Film panel open over the collection, its saved articles sinking like keys when pressed. — 4.1History bloom, Saved keycap, home panel sequence

chapter 05

reflection

fork: a coachmark on home or teaching it once at selection
chose: copy at selection, and the minimum raised to three
cost: the confirm screen, cut after five rounds
gain: one clear contract, taught once

Fifteen testers picked four to six interests, got three reads, and asked where the rest went. I taught the split at selection.

Domain selection with five picks: “5 selected · first three lead · two broaden”. — 5.1selection copy at the moment of choosing

Philo is a UI and interaction study, April to July 2026.

If you want to talk about the work, write me at cfay.work@gmail.com

Next project: rebecca

philo

interactive prototype

three articles. no infinite feed.

focused content · expressive interaction

12s · tap to preview

Open Philo

about philo

Philo builds a daily edition of three articles from the domains you choose. Expand a panel, open a story, and save what you want to return to.

application · case study

philo

role: product designer
format: mobile web prototype
implementation: AI-assisted prototype
span: April 2026 – July 2026

context

a finite daily edition built from the reader’s chosen domains

limit

backend, production readiness and content licensing not established

thesis

Focused content.
Expressive interaction.

chapter 01

premise

framing, before: news curation
research method: discovery interviews
observation: “news” was associated with politics and negativity; the frustration described was wider than news
framing, after: interest curation

In discovery interviews, people associated “news” with politics and negativity. The frustration they described was broader: reaching material they cared about without sorting through the rest of the web.

The interviews informed the shift from news curation to interest curation. They did not specify a solution.

Limiting each day to three Articles is a design decision, made against an internet that maximizes available content.

chapter 02

curation

step 01 · domains: three to six of twelve; the first three lead the edition
step 02 · focus: interests within each chosen domain
step 03 · refine: optional, in the reader’s own words
testing observation: several people rushed onboarding or missed the deeper curation controls
design intent: interests sharpen what each domain contributes
current prototype: the edition is built from domains alone; focus choices are stored and not yet used; refine text is not kept

Philo asks; it does not watch. The reader states what they care about in three steps: Domains, then Focus within each Domain, then an optional Refine. Declared intention over inferred behavior: the prototype records nothing about what is read and changes no preference on the reader’s behalf.

Post-prototype testing found a weakness in the order. Several people rushed through onboarding or missed the deeper curation controls. Figure 2.1 shows why: Build My Edition becomes available at the third Domain, in a bar fixed to the bottom of the screen, while Focus and Refine sit below the fold.

A three-Domain minimum was added after testing, so that every Edition has its three lead Domains. It protected the Edition’s structure. It did not solve the larger issue: the setup asks for attention before the product has given anything back.

Onboarding after three domains are chosen. Build My Edition is active; Focus is below the fold. — 2.1 onboarding, in order · one scroll, three steps

limit

Three Articles.
No infinite feed.

chapter 03

daily edition

articles per edition: three · a design hypothesis, not a research finding
three, in the prototype: fixed in four places: edition slots, lead domains, accent colours, the counter
to open an article: two steps: expand a panel, then open the article
expand · collapse: 700 ms · 500 ms
motion rule: dissolve in, cut out
testing evidence claimed: none, for how the interaction feels

Three Articles per Edition is a design hypothesis. One felt brittle. Two stayed restrictive, without enough variation. Three gives enough choice for one recommendation to miss while the Edition stays finite, and it suits the accordion.

The accordion makes choosing a two-step act: a panel expands, then the Article opens. The friction is deliberate. That is design intent, not a tested result; no testing evidence is claimed here for how it feels.

Figure 3.1 is the expansion itself, from the running prototype: each frame is the prototype’s own transitions held at the stated time. The score beside it lists what moves and when, read from the source. Content arrives by dissolve and scale. On collapse it is cut, not faded.

The Daily Edition at the start of the Film panel expansion, 0 milliseconds. — 3.1the prototype’s transitions

The Daily Edition at rest, with three equal panels for Film, Art and Music. — 3.2the three states of choosing

chapter 04

saved

layout: by the number of domains holding a saved article, one to six
position: by rank of saved count
row height: by share of saved articles, between 100 and 410 px
colour: by the domain’s priority, not its position
exception: the one- and four-domain layouts are fixed and do not follow distribution

Saved is not a list. Each Domain holding a saved Article takes one cell, and four rules place it.

Figure 4.1 is one screen in three states: three Domains saved evenly, six Domains, and three Domains saved unevenly. Only the chosen Domains and the saved Articles differ between them; the states were set through the prototype’s own data.

Saved articles distributed evenly across Film, Art and Music, two in each domain. — 4.1one rule set, three states

chapter 05

reading · reflection

reader: an overlay over the edition; opens in 420 ms
leaving: the home control, or a downward pull from the top of the article
position mark: a dot on a 64 px rail: position within the article; nothing is stored
the room: a design principle; not tested as a finding
history: unresolved

A screen should feel like a bounded place: the reader knows where they are, enters and leaves deliberately, and unrelated destinations do not run together into one accidental scroll. This is a design principle. It was not tested as a finding.

The reader is where it is clearest. The Article opens over the Edition and closes back to it. The mark at the right edge reports position inside a finite Article and nothing else; it stores no progress.

History is unresolved. Days run horizontally and Articles vertically, and the arrangement does not hold across different amounts of content: a day with few Articles leaves the vertical dimension underused. It also shows wider drift in the visual system across Philo’s secondary screens.

A next iteration would shorten the first-run setup, make progression and scroll cues clearer, and show how each choice changes the Edition as it is made. It would also set shared rules for the secondary screens before adding to them.

A Film article opened over the Daily Edition. — 5.1the reader, and where coherence ran out