Internal · HOOPT team · September 2026

The app looked in the mirror.

Six screen recordings, every main tab plus a brand-new user, analyzed frame by frame. This is what the live app does brilliantly, what breaks the story, and exactly what we fix before we go to market.

6recordings audited
54:24minutes of footage
14.3klive items, zero stutter
23tracked issues
1high severity
0story blockers

Navigate with the chips above, or ← → between chapters.

Why this audit exists

The call: we launch with the live app.

Baja stays our momentum engine — all four releases keep landing on schedule. But the go-to-market push runs on the app people can download today: 6.2.1. So the live app has to carry the story on its own, to a stranger, with no founder standing next to it.

That framing changes what matters. A missing feature is fine — Baja is coming. A broken promise is not: our pitch is honest numbers, private by design, and effortless organization. Anywhere the live app contradicts that pitch — a counter that lies, copy that drifts, a superpower nobody can find — is a launch item, not a backlog item.

The verdict up front: the foundation is genuinely strong. Performance is flawless across 54 minutes of footage, the intent grammar holds across every surface, and the differentiators are real and demonstrable. The work before launch is precision, not surgery — one search fix, a handful of counters and words, and unhiding our own best features.

Six tabs, six grades

Scorecard

Graded on one question: can this surface carry a new user, alone, at launch?

A−CollectionsA 14.3k-item library that never stutters — best feature hidden behind an unmarked door. A−ConnectionsChat that compounds into a library. One pill everyone will tap doesn't work. B+Search & DiscoveryGrade-A relevance and honest ranking — pointed at the wrong continent. B+Recent ActivityThe feed with nothing to sell you. Two length problems and a counter that lies. A−ProfileIdentity without an audience — and 492 albums in one endless column. B+Onboarding2½ minutes to a working Life OS — powered by a founder who can't welcome everyone personally.

Chapter 1 · Video 1 · 7:58

Collections

A−

A 14.3k-item library that never stutters — with its best feature hidden behind an unmarked door.

v6.2.1 · 11.7k places, 1.7k movies, 847 books · 67.3k items across the network

Loved it — keep exactly this

  • Zero-input rich architecture. One tap of Wanna or Love and the item arrives fully dressed — photos, hours, ratings, streaming, page counts. No forms, ever. 04:19
  • Real-time facet queries. Netflix + Documentary recalculates to "SEE 42 / 1.7K MOVIES" live as chips toggle. The database feels alive. 05:11
  • Smart Navigation. Filter to Hermann Hesse, tap the header, and see exactly which friends share that shelf — then step straight into Anthony Howard's three Hesse books. Our single most magical interaction. 06:39–07:07
  • Navigation discipline. Modals dismiss with ✕, pushes return with ← , and every scroll position survives the round trip. Zero state loss across the whole recording.

Wanna fix

  • "Select List" undersells a query engine. The label promises a folder picker; it opens our multi-facet filter engine. Power users will never find it. Medium ISS-01
  • Smart Nav is invisible. The header-chevron context switch has no affordance cue at all — the audit calls it "completely undiscoverable without trial." Medium ISS-02
  • "My Theater" reads as showtimes before first tap; Atlas / Theater / Library mix three metaphors. Low ISS-03
  • Carousels rely on edge-peek alone — no dots, no scroll cue. Low ISS-04

Today — the label promises a folder picker

✗
Select List ⌵hides the multi-facet query engine behind four dead words

Candidate relabel (ISS-01, test with users)

✓
Filters & Lists ⌵"SEE 42 / 1.7K MOVIES" finally gets found

The pattern starts here

Our two most impressive capabilities on this tab — the facet engine and Smart Nav — are the two things a user cannot discover. Both got flagged "understandable only after interaction." This repeats across the app; it becomes Pattern 01.

The recording can't answer: how a user adds an item from this tab, what empty categories look like for a day-one user, and how logged items get edited or deleted. All three matter at launch — the empty-state question most of all.

Chapter 2 · Video 2 · 10:18

Connections

A−

Chat that compounds into a library. One pill everyone will tap doesn't work.

Direct chat · Common Items ∞ 133 · Community Collections 22.5k items · AI community creation

Loved it — keep exactly this

  • The Common Items engine. ∞ 133 shared items with James, auto-split into Love and Wanna — an objective map of a friendship, computed with zero effort. 01:13
  • Community Collections. A 22.5k-item group library organized by city and genre, with member attribution and the privacy shield on every sheet. Chat that never evaporates. 05:27–06:29
  • AI community creation. Name + 12-word description → custom cover art in 4.8 seconds → confetti → live chat. Six taps, zero friction, real delight. 08:28–09:57
  • Timezone-aware calling. "It's 04:45 PM in 🇺🇸 Los Angeles where James is currently located" before you ring. Intentionality as a feature. 01:54

Wanna fix

  • "132 in Common" isn't tappable. It looks like a button, sits in the header, and is the doorway to our best social feature — but you have to go through Friend Details instead. Everyone will tap it. Medium ISS-05
  • The 9+ badge has no breakdown — direct messages and community noise weigh the same. Low
  • Privacy reassurance sits below the fold on member attribution sheets — the one line that earns trust can be missed. Low ISS-08
  • AI cover art can't be cropped or styled; take it or regenerate. Low ISS-06

Today — the header pill looks tappable and isn't

✗
James Orozco132 in Commontap → nothing

After ISS-05 — the pill is the door

✓
James Orozco132 in Commontap → Common Items

One-line fix, outsized return

Making the in-common pill tappable is probably the cheapest ticket in this entire audit — and it unlocks the feature that best demonstrates what HOOPT is in a demo, a review, or a friend showing a friend.

Still to verify: what the Notes tab in Common Items actually contains, how non-admins see Community Details, and what an invite link looks like to someone opening it on desktop or Android — that last one is a GTM-path question.

Chapter 4 · Video 4 · 9:22

Recent Activity

B+

The feed with nothing to sell you. Two length problems and a counter that lies.

Strict reverse-chronological · zero ads · zero public metrics · 5:00 HD video · 60fps throughout

Loved it — keep exactly this

  • The purest feed in the industry. Timestamps run 1h → 4h → 5h → 6h → … → 2d with zero reordering, zero insertions, zero engagement bait. Verifiable on screen. §6
  • The skydiving loop. 5-minute HD video with sound → album Fly Days 🪂 → Santa Cruz city page → Wanna Go. A friend's lived moment becomes your future intention in three taps. 05:04–06:11
  • Smart batch tagging. North Beach applied to 7 photos in two taps — Select All → Add to Selected Media. Life logging without the labor. 00:52
  • Replies go private. Tap ↩ and you're in a 1-on-1 composer with the life object pinned on top. No public comment theater, ever. 02:47, 06:35

Wanna fix

  • Reactions inbox repeats the full card per clap. One skydiving photo, four reactors, four identical cards. Group as "Max, Emre +2 clapped." Medium ISS-09
  • Show More +80 opens an 80-row wall inline — 15+ swipes to get past one friend's movie binge. Cap around 10, open the rest in a sub-view. Medium ISS-11
  • The reactions counter said "• 0" above an inbox holding dozens of claps. In an app whose brand is honest numbers, a lying counter on the home screen is a brand bug wearing a Low-severity costume. Fix as Must ISS-10
  • Abandoned replies vanish — no draft, no prompt. Low

Today — the counter lies, and every clap repeats the full card

✗
Recent Activity• 0 ›
the inbox behind it holds dozens of claps
✗
Max clapped at your video 🎬 — Santa Cruz
Ertuğrul clapped at your video 🎬 — Santa Cruz
Emre clapped at your video 🎬 — Santa Cruz
one skydiving video, three identical cards

After ISS-09 · ISS-10 — a true badge, one grouped card

✓
Recent Activity• 12 ›
✓
Max, Ertuğrul +1 clapped at your video 🎬

The claim we can prove on camera

"No ads, no follower counts, no public likes, no algorithm" is fully verifiable in this one recording. That makes Recent Activity our strongest live demo surface for press and investors — provided the counter tells the truth first.

Chapter 5 · Video 5 · 7:17

Profile

A−

Identity without an audience — and 492 albums in one endless column.

14.3k collected · 2.6k friends · 580 communities · 240 quotes · 492 albums, 27,583 media items

Loved it — keep exactly this

  • No follower model, anywhere. No likes, views, reach, or impressions on any surface. Identity is what you've lived and collected, not how you performed. The audit confirmed the absence across every screen. §4
  • Three lives, one carousel. Currently in SF · Where I live · From Istanbul — swipeable map cards that describe a person the way people actually describe themselves. 02:07
  • The wisdom layer. 240 quotes across Personal Insights, Iconic Quotes, Gems from Friends — "each quote is a window into your mind, guiding your AI engine." Depth no competitor's profile has. 02:32
  • Live localization. English ↔ Türkçe swaps the entire interface in ~1 second, cleanly. Ready for a two-market launch out of the box. 00:46–01:12

Wanna fix

  • 492 albums, one unsectioned grid. The Life Gallery is our richest surface and the hardest to scan — no sections, no filters, just scroll. Medium QA-PROF-03
  • 14.3k collected vs 27,583 items — the two headline numbers on one screen with no visible relationship. Users will do the math and get confused. Define
  • Settings resets your scroll back to the top of Profile on return. Low QA-PROF-01
  • The support bubble squats on content bottom-right; should auto-hide on scroll. Low QA-PROF-02
  • No data export observed in Settings — for a "your data, your life" product, that absence will get noticed. Check

Today — 492 albums, one endless two-column scroll

✗
Morning Rituals · Fly Days · Grocery Shopping · 2026 Hikes · Kilito Maintenance · SF Restaurants · …486 more

After QA-PROF-03 — sections, pins, search

✓
⭐ PinnedSF Restaurants · Fly Days 🪂
RecentMorning Rituals 🥪 · 2026 Hikes
⌕ Search albums

The audit's own words

"Identity is expressed through structured life context rather than audience performance." That sentence is GTM copy. The profile already proves it — we just need the gallery to scale with the life it holds.

Chapter 6 · Video 6 · 10:37 · new account: @celinedior

Onboarding

B+

Two and a half minutes to a working Life OS — powered by a founder who can't personally welcome everyone.

Invite code → account → 5 items in 2:27 → HOOPTer #6,124 → auto-organized Collections → 4 common items with the inviter

Loved it — keep exactly this

  • The 5-item calibration works. Two movies, two books, one place, ~2½ minutes — then Collections lands already organized with genres and directors. Instant proof of the whole promise. 02:23–05:08
  • The taste graph fires immediately. Five seed items in, and Celine already shares 4 common items with Ali — computed, not inferred. Our differentiator, demonstrated in minute five of membership. 07:51
  • HOOPTer #6,124. A membership number instead of a follower count. Belonging as the first reward. 04:57
  • Money talk in the open. "How does HOOPT make money?" one tap from the welcome thread, with real prices. Location permission asked contextually, with a real reason — graded A by the audit. 07:09, 03:03

Wanna fix

  • The warmth doesn't scale. Celine's cold start felt alive because her inviter was Ali — active feed, rich communities, personal welcome. An inactive inviter yields a near-empty Connections tab and a silent Activity feed. The push will mint exactly those users. Strategic
  • A modal interrupts the flow at 3/5 even though the bottom banner already tracks progress. Medium QA-ONB-01
  • 350M+ on the carousel, 300m+ on the milestone card. Same catalog, two numbers, ten screens apart. Low QA-ONB-03
  • "Collect 5 things you love" — but Wanna counts too. Say what we mean: love or want to try. Low QA-ONB-04
  • "Mosaics" appears on screen one and never again. Define it in-app or cut it from the carousel. Low
  • Ali's communities: 580 on his device, 257 from Celine's view. Probably privacy filtering — but verify, and make sure it's intentional. Verify QA-ONB-05

Today — the same catalog, two numbers, ten screens apart

✗
Carousel 2:"350M+ places, movies & books"
✗
Milestone card:"300m+ places, movies, and books"

After the copy canon — one figure everywhere, owned by Ece

✓
One canonical numbercarousel · milestone · web · App Store

The one strategic finding of the whole audit

Onboarding's magic is currently a person, not a product. Before the push, the founder-welcome experience — the thread, the Starter Kit, community previews, the first-day guidance — needs to be systemized so every new user gets Celine's day one, whoever invited them.

What repeats across all six videos

Four patterns

Individual tickets are listed per chapter. These are the shapes they make together — fix the pattern and half the tickets close themselves.

01

Hidden superpowers

Smart Navigation, the Common Items pill, the Select List query engine — our three most demo-worthy capabilities are the three things a user cannot discover. Every one was flagged "undiscoverable without trial" or literally untappable. We built the magic and then hid the doors. Labeling, affordance, and one working pill fix this pattern almost entirely.

02

Counter drift

Our entire pitch rests on numbers people can trust — the ∞ counts are the product. Yet the reactions badge read • 0 over a full inbox, Ali's communities read 580 on one device and 257 on another, and 14.3k collected sits beside 27,583 items with no stated relationship. Each is small; together they erode the one thing we can't afford to erode. Rule for launch: every visible number is true, or it isn't visible.

03

One copy canon

350M+ vs 300m+. "Things you love" vs a flow that counts Wanna. "Mosaics" defined nowhere. "Select List" describing a filter engine. "My Theater" sounding like showtimes. None of these is a bug; all of them are the same bug — copy written per-screen instead of from one canon. One afternoon with Ece and a single source-of-truth sheet closes the whole pattern.

04

The inviter is the product

Two of five tabs (Connections, Recent Activity) draw their day-one value entirely from the inviter's activity. With an Ali-grade inviter, HOOPT feels alive in minutes; with a dormant one, half the app is an empty room. A GTM push recruits strangers at scale — the product has to supply the warmth the founder currently supplies by hand.

The core deliverable

The list

Everything above, compressed into one ordered plan. Must ships before the push. Should rides the next release. Polish fills sprint gaps. Cross out as we land them.

Must — before the push

the launch gate · 8 items
  1. Local intent wins generic searchWeight the user's geography above global popularity for generic queries — "chicken fried" in SF returns SF first. The audit's only High, on the most common query type there is. QA-SRC-01
  2. Every visible counter tells the truthSync the reactions badge (• 0 over a full inbox), verify the 580 / 257 communities split is intentional privacy filtering, and reconcile Collected vs Life Gallery counts in copy. Honest numbers are the brand. ISS-10 · QA-ONB-05 · QA-PROF
  3. Make "N in Common" tappableThe header pill opens Common Items directly. Cheapest fix in the audit; unlocks our best demo feature. ISS-05
  4. One copy canonSingle catalog number everywhere (350M or 300M — pick one). "Collect 5 things you love or want to try." Define Mosaics in-app or cut it from the carousel. One sheet, owned by Ece, checked before every release. QA-ONB-03 · 04 · Mosaics
  5. Kill the 3-of-5 interstitialThe bottom banner already guides activation; the modal only interrupts a user who is mid-exploration — exactly the wrong moment. QA-ONB-01
  6. Group the reactions inboxOne media card per item with a reactor avatar stack — "Max, Emre +2 clapped at your video" — instead of four duplicate cards. ISS-09
  7. Cap inline feed expansionShow More reveals ~10 rows inline; the full 83 open in a dedicated view. Protects the feed's rhythm from one friend's binge. ISS-11
  8. Systemize the welcomeProduct-carried version of the founder cold start: automated welcome thread, Starter Kit, curated community previews, and inactive-inviter scaffolding — so every push-recruited user gets Celine's day one. The audit's one strategic finding.

Should — next release

fast follows · 6 items
  1. Swap the notes glyph≡ reads as a drag handle in two tabs; a note or bubble glyph says what it is. QA-SRC-03
  2. Relabel "Select List"Test "Filters & Lists" (or similar) so the query engine sells itself. ISS-01
  3. Give Smart Nav a doorA visible affordance on filtered-view headers, beyond the one onboarding tooltip. ISS-02
  4. Section the Life GalleryGroups, pins, or search across 492 albums — the surface should scale with the life it holds. QA-PROF-03
  5. Pre-permission photo cardOne branded line — "pick your profile picture" — before iOS discloses a 27k-photo library. QA-ONB-02
  6. Draft-save Add to My Life & repliesAn accidental dismiss shouldn't discard seven tagged photos or a half-written reply.

Polish — when sprints allow

4 items
  1. Preserve Profile scrollReturning from Settings keeps your place. QA-PROF-01
  2. Auto-hide the support bubbleSlide away on downward scroll. QA-PROF-02
  3. Prefetch the pausesFirst album thumbnails, city-list items, map pin clusters — erase the 0.4s skeletons. PERF-03 · QA-PROF-04
  4. AI cover art controlsCrop / reposition, and style presets on regenerate. ISS-06

Do not touch

Protected species

The audit independently confirmed each of these on camera. They are the moat — no redesign, no "improvement," no A/B test gets near them.

The honest feedStrict reverse-chronological, zero insertions, zero reordering — verifiable by reading the timestamps on screen.
Zero vanity metricsNo follower counts, no public likes, no view tallies, anywhere. Every reaction is a named friend.
Monotonic ranking333 → 296 → 238 → 171 → 156. Consensus order, no sponsored slots. This is the pitch, rendered in UI.
Private repliesComments route into 1-on-1 conversations with the life object attached. No public comment theater.
No ads, everSix videos, 54 minutes, zero banners, zero sponsored rows, zero promoted anything.
The performance bar60fps through 14.3k items, 83-row expansions, and 5-minute HD video. Not one dropped frame in the footage.

For the go-to-market push

Ammo

Claims the audit marks safe to use today — each one demonstrated on camera and provable in a live demo.

Search rankings follow human consensus, in order, with zero ads and zero paid placement.Proven: Kokkari case study, Video 3 — perfectly monotonic ∞ counts across the top five.
HOOPT has no followers. Friends are two-way relationships, and no surface shows likes, views, or reach.Proven: Profile audit, Video 5 — absence confirmed across every screen.
Add something you love and HOOPT dresses it for you — photos, hours, ratings, streaming — with zero data entry.Proven: Collections audit, Video 1 — the zero-input architecture.
Five items and about two minutes gets you a working, auto-organized personal Life OS.Proven: Onboarding audit, Video 6 — 2:27 elapsed, Collections pre-sorted on arrival.
Your feed is your friends' real lives in the order they lived them — including five-minute videos, not fifteen-second bait.Proven: Recent Activity audit, Video 4 — strict reverse-chron, long-form HD with sound.
The moment you join, HOOPT shows you what you and your people already have in common.Proven: Videos 2 & 6 — ∞ 133 with James; 4 common items for a five-minute-old account.

Hold until verified: "the only unmanipulated collective intelligence" (needs comparative audit) · end-to-end encryption specifics (needs cryptographic verification) · Premium $20 / Pro $200 availability (needs billing check) · "search intent feeds the collective intelligence" (needs privacy-policy alignment). Strong claims — earn them before we print them.