4f. Practice Deck (actions, not audio) - forming at v0

Affirmations to speak, journaling prompts, tapping how-tos, manifesting techniques, micro-practices. Chart + context personalized, refreshed by the Mindset Ninja agent (North Star 6b) as she comes online; v1 can be a static personalized set shaped from the Understanding. Each card: do-it-now body, "why this, for you" line, and a done-tap (counts toward constellation fills where relevant).

4g. Voice Lab - Wave 2 (cheap, high-signal)

Jeff's voices section: - Audition shelf: named candidate voices (Fish + ElevenLabs; the registry is Affirmology_VoiceRegistry_v1.md), each with a 20-30s standard sample rendered ONCE from a fixed house script (not per person). Users: play, speed-toggle, thumbs up/down + "would you want your readings in this voice?" Kills duds fast, feeds the voice roster. - Per-person voice preferences: chosen voice per audio TYPE (female soul song voice, male readings voice, etc.) saved to profile, honored at render. - SPANISH shelf: 2-3 Spanish candidates (Fish s2 handles Spanish; test first), same audition flow. First expansion language; our tester base skews Spanish-speaking. - OWN-VOICE experiment (flagged, later): Fish voice-clone from a reading sample, one test audio in their own voice. Consent copy + expectations ("some people love it, some hate it") + QC gate. Run as an A/B experiment with a handful of volunteers, not a default feature.

4h. Feedback Hub - v0 LIVE (it is mostly copy + a text/voice box)

• THREE BOXES, per Sol's 7/11 voice note (feedback means ALL kinds of things, not just fixes): 1. Ideas for the future ("anything you wish this could do"). 2. What didn't land ("anything you didn't like or didn't resonate with"). 3. Free write (open paragraph form, no structure, whatever you'd like). Each accepts text or a voice note (transcript-only rule). Tagged on submit so triage stays easy.
• A short "How to give great feedback" card sits at the top of the Hub (Cowork copy, from Sol's guidance): what we love hearing about (the voice, the music, the structure, the delivery, breakthroughs, and anything else), and that ideas count as feedback too.
• Your feedback log: what you said, per audio, and WHAT CHANGED because of it ("You said the pacing rushed. v2 of your Night Install is re-paced."). Closing the loop is the retention magic and the reason testers keep giving gold.
• A/B invitations appear here (and via Oracle): "You've had version A of your Morning Claim. Here's B. Which one is more you?"
• Feature requests: a lightweight "I wish the app..." lane, so ideas stop going to Jeff's texts. Intelligent triage backend later (7.3); capture now.

4i. Profile + Settings - v0 LIVE (thin)

• Birth data (view; edits gated through support for chart integrity), FULL BIRTH NAME field + "name you go by" field (both feed numerology/nameology; Jeff's ask), email, voice prefs, speed prefs (auto-learned, viewable), notification prefs.
• Birth-time helper (the Randy problem): if birth time was defaulted (noon) or marked unsure: (1) their initial suite AVOIDS time-sensitive structures (no rising-sign or house-heavy content; the engine flags which structures are time-sensitive), with an honest note; (2) a guided rectification flow: adaptive A/B questions walking the time window down (morning or night → window → "are you more X or Y" keyed to rising-sign/house contrasts), run through the agent council (the Olga precedent, Affirmology_Olga_Rectification_v1.md). Wave 2-3 build; the AVOIDANCE rule is Wave 1 (it is a render-side filter, cheap).
• Engagement stats ("your listening", streaks in identity language, per Habit Engine).

4j. Help / Get Started - v0 LIVE (pure copy, Cowork writes it)

"How to use this app well": the daily rhythm, how to talk to your guide (including voice notes), what the constellation is, how feedback shapes your audios, how to request new audios. Text first, short videos slot in later (the wiki/walkthrough assets from the weekend plan feed this). Also the support path (Hermes voice → hello@affirmology.ai, already built in chat).

5. THE CONSTELLATION SYSTEM v1 (the gamified map)

Seeded from the Gamification doc + North Star section 4; this section makes it buildable.

5.1 Principles (locked from the docs)

• The constellation is a PROGRESS WALL (Jeff, 7/10): visible, unmissable, accumulating micro-wins, a healthy challenge paired with the real mechanistic because. It is SCAFFOLD around being seen, never the pull itself.
• ACCRUAL, NEVER LOSS (locked guardrail): constellations fill up; they do not break. Copy says "You have been aligning for twelve days", never "don't lose your streak". No broken-flame anything.
• The constellation REFLECTS practice; it never GATES the daily practice. Nobody is ever locked out of their doses. Unlocks add bonuses (guides, deep sessions, ceremonies), never withhold basics.
• Identity language: not "7-day streak" but "7 days living as her" (their chart language).

5.2 Constellation Zero: ARIES, The Initiation (v0 build)

Jeff's spec, refined: - 5 stars = the 4 daily moments (morning, meditation, movement, before sleep) + the weekly reading. - Each star fills in 7 fractions; one fraction per day that slot is completed (max one/day/slot). Misses don't reset; the star holds its fill (building the habit, not punishing). - When a star completes it IGNITES (bright, named). When two ADJACENT stars are lit, the line between them draws itself. All 5 lit + lines = the full figure of the Ram fades in around them (the constellation art moment) + the completion ceremony. - Completion reward (per the Gamification doc's headline mechanic): ACTIVATE YOUR GUIDE. Completing Aries unlocks choosing a Wisdom Council member as their personal guide (the customer-facing voice Sophia/Hermes bring forward). This drives base-audio consumption, makes the guide feel earned, and prevents day-one choice overload. Plus: their week-in-review "release notes for the new you" (which doubles as the day-7 feedback ask + the SOS v2.0 version bump). - Why Aries first: first of the zodiac, action + initiation. On completion the sky rotates and the other 11 shapes become visible territory.

5.3 The sky (the map UI)

• A pannable/rotatable star field (globe-feel). v0 can be a 2D parallax pan, not true 3D; ship the feeling, upgrade the tech later. The 12 gold zodiac SVGs already exist (zodiac/ asset set).
• States per constellation: FILLING (current), AVAILABLE (tappable, shows its rules + what it unlocks), FORMING (visible shape, grayed, "not yet available"), HIDDEN (deep-sky ones that appear only when conditions are met).
• Tap any constellation → its card: theme, rules, star list, reward, progress.

5.4 The constellation taxonomy (each journey type = a constellation family)

• THE 12 ZODIAC = the base tier (lowest membership level). Each is a themed program: Aries = initiation/habit; Taurus = prosperity + body; Gemini = voice + expression; Cancer = family + home; Leo = confidence + being seen; Virgo = craft + daily alignment; Libra = relationships; Scorpio = shadow + rewire; Sagittarius = meaning + expansion; Capricorn = vocation + mastery; Aquarius = your genius + community; Pisces = dreams + sleep + spirit. (Map journeys from 4c onto these; the themes flex per person.)
• SYSTEM CONSTELLATIONS (beyond the 12, can be paid/deeper tier): the Gene Keys Deep Journey (stars = Activation mastered, Venus mastered, Pearl mastered, the woven journey, a reading set), Human Design, Vedic, Numerology. Same mechanic, system-specific stars.
• CONTEXT CONSTELLATIONS: wealth, relationship healing, IFS/parts, depth-psychology (deep-survey gated), the long Hero's Quest (Jeff's Ophiuchus instinct is right: give the epic journeys a serpent-bearer constellation of their own, off the main wheel).
• HARD-MODE constellations (later): e.g. 3 audios/day for a month, strict rules, big unlock. Never the first two; earn trust before asking for commitment mechanics.
• Multiple constellations fill IN PARALLEL (locked in the Gamification doc); sequence is only forced for Aries → everything else.

5.5 Unlock economy (rewards for completion, from the docs + Jeff's list)

New guide activation (the headline), a deeper reading, a ceremony audio, chatbot create-credits, early access to new structures, a shareable constellation badge (5.6), bonus deep-journey sessions. One reward table lives with the constellation config so it is data, not code.

5.6 Shareables (the social loop, later wave but design the hooks now)

THE CURIOSITY GAP DOCTRINE governs (Sol, locked in the brand synthesis): we never publish the audio, we publish the REACTION. So shareables are: the branded constellation card (their constellation, their name, one line from their chart), the day-7 release notes, reaction moments. NO audio snippets in shares, ever; the product stays unheard and the only way to hear it is to get your own. Pride for them, intrigue + a personal link for the friend (ties into 4e's loop).

5.7 Asset production (Jeff + Higgsfield/Nano Banana; instructions to generate now)

Needed per zodiac constellation (start with Aries only): 1. Star-field base tile (deep emerald #0a2014, faint nebula, our palette; seamless pan). 2. The constellation in 4 states: dim points → stars igniting (per-star glow sprite) → gold lines drawing → the full mythic figure faded in behind (marble-and-gold Hellenistic style matching the council busts; the existing zodiac/ SVGs are the line-work base layer). 3. A star-ignition micro-animation (one reusable sprite/lottie) + a completion ceremony moment. Prompt style: reuse the locked visual language (emerald/gold/cream, Cormorant, painterly constellation figures per the Affirmology_GraphicsLibrary_Handoff_v1.md system). Static PNGs + one sprite is enough for v0; video flourishes later.

6. THE LEARNING BACKEND (feedback intelligence; extends Map v2 sections 3-4, not new)

• Instruments: exactly the Map v2 testing kit (2-tap, generic↔me slider added, block intensity d0/d7, voice notes, behavioral). Nothing else at Wave 1.
• Per-person preference store (new, small): learned settings per audio TYPE: playback rate, pause pref, voice pref, length pref, bed pref. Written by feedback events, read at render + play time.
• The A/B distributor: a backend job that (a) assigns variants per Map v2's arms, (b) tells the app which card to label "try version B", (c) collects the paired outcome. One experiment per person at a time (do not stack tests on one listener; it muddies signal and trust).
• The decision rule stays Map v2's: settled-universal issues auto-adopt; contentious ones gather more data. Weekly digest to Jeff (email or the briefing chat) instead of a live dashboard for now.
• Technique tracking: already solved by build reports naming techniques; ratings map back through audio ID → build report → craft entries → confidence re-weighting (Assimilation step 5, the small unbuilt piece; DATA owns).
• Structure persistence rule (Jeff's ask): when a person loves a structure (high landed + high "deeply me" + replays), the adaptive default REUSES that structure family for new contexts before experimenting again; "explore new techniques" becomes an explicit opt-in card, not a surprise.

7. CONTEXT, THE BELIEF BOX, AND COST

7.1 The one-context pick (onboarding, v0-adjacent)

This IS the three-move product story from the brand work: pick your context, get your suite, listen and align. At entry (or day 2): "What are you actually working on right now?" ONE pick: Wealth / Love + relationships / Purpose + work / Health + energy / Peace + confidence. The initial suite is then rendered with that context woven in (the context layer the beta group needs to feel: they already know their chart; context + transformation is what will impress them). Week 2: the Oracle refines it in conversation. - THE ASK COPY (approved, use verbatim at the first habit prompt): "We ask you to listen every day, because becoming is not a decision. It is a repetition. One listen will move you. Thirty is when you stop having to try." (The because must stay mechanistic; the Langer research says a large ask needs a real reason.) - THE LAYERS FRAMING (critical, Jeff 7/10): layers 2 and 3 (transformation tools + context) are never presented as unbuilt. The line is: "We have not added your life. Not because we can't, because you have not told us yet." Every context ask, discovery conversation, and upgrade moment is an invitation to be known more fully, never a paywall pitch.

7.2 The Limiting Belief box (Wave 2)

One named block, in their words (this already exists at intake per Map v2's block-intensity metric; this makes it a visible object in Explore). The app SHOWS which audios will respond to it: "Your belief is being worked in: tonight's Night Install (reconsolidation pass), your EFT micro, your visioning." When /render-patch (2.2) exists, a belief update can patch one line into an existing loved audio for pennies. The block-intensity 1-10 at d0 and d7 measures it. One block at a time.

7.3 Feature-request intelligence (later)

The 4h request lane feeds a weekly triage pass (agent): cluster, compare against roadmap + vision, surface the 3 that matter with who-asked and why. Never auto-promise anything in-app.

7.4 Cost discipline (how this stays affordable)

• Client-side rate control instead of re-renders. /render-patch instead of full re-renders.
• Written blueprint sections: one cached formatting pass from the stored Understanding (pennies, once) instead of council runs. Chart SVGs: free (already computed).
• Request-desk renders are allowance-based (N new audios/week at beta; generous but visible), and Fish voices carry the volume (the grant application is in flight).
• The council runs once per person (the Understanding), everything else shapes from it. Unchanged.

8. WHAT JEFF ASKED THAT IS ALREADY ANSWERED (fast corrections so nothing gets re-researched)

1. Voice-note tech = Parakeet on-device (sherpa-onnx) with Whisper interim, NOT ElevenLabs. Wispr is your personal dictation app, unrelated. (PROJECT_STATE 2026-06-18 correction.)
2. Speed control = client playback rate, not re-rendering. Pause width = next-render setting.
3. The Explore/Listen/Oracle 3-tab call was already logged with Code on 7/10; this doc specs it.
4. Chart graphics exist in code (pdf_chart_svg.py); the app work is one endpoint + one screen.
5. The relationship engine (incl. Penta) is built; the app work is surfacing + the invite loop.
6. Constellation-unlocks-guide mechanic was locked 6/20 in the Gamification doc; kept here as the Aries completion reward.
7. "Which system am I leaving out": none; the five are all listed. The missing VIEW was the five-system synthesis diagram, which is also already built.

9. AUDIT: PUSHBACK + OPPORTUNITIES (the challenge section Jeff asked for)

1. Protect the floor. The single biggest risk this weekend is Explore eating the Daily Pair. Map v2's floor (Daily Pair + feedback taps) beats every menu. Explore v0 is deliberately mostly forming-cards. Resist making more of it live before Monday.
2. Do not put five dials at audio end. Two taps + optional expander. The completion moment is a celebration; burying it in instrumentation costs the habit you are building. Deep dials go to the Hub and the Oracle's occasional asks.
3. The Oracle top-bar: keep it out. Everything you listed for it (histories, reports, roster, routines) has a natural Explore home. A pure chat surface is rarer and more premium than another mini-nav, and the starters give it proactive reach without chrome.
4. Constellations: reflect, never gate. The moment a star withholds a person's Night Install you have a dark pattern and it violates the locked ethics line. Bonuses only.
5. A/B etiquette: one live experiment per person. The beta group is 30-some people; spreading 6 experiments across them yields noise. Sequence tests, per Map v2's one-variable rule.
6. Own-voice and dance-track ideas are gold but Wave 3. Both are novel-render-pipeline work (clone consent + music generation licensing). Park them visibly in the Voice Lab / Journeys forming cards so they are promised, not built.
7. Missed opportunity you did not mention: the Anchor Word. It is the carry system's spine (Map v2) and it wants a UI home: their word, quietly visible on the Listen header or their Blueprint page, tappable to replay its install micro. Cheap, deeply on-brand.
8. Missed: the week-in-review as the share moment. You asked for shareables; the day-7 "release notes for the new you" is the natural first shareable (badge + one chart line), better than sharing raw audio clips, and it is already in the Habit Engine spec.
9. Missed: no-birth-time users are a Monday-relevant content rule, not just a future feature. Randy is IN the cohort with noon-default data. The render-side filter (avoid rising/house-heavy structures for unsure-time profiles) should be in the Wave 1 render configs.
10. Two metaphors, one app (reconcile deliberately). The brand work approved the DAILY MEDICINE frame ("your daily subconscious supplements", the dose language) as the product's organizing metaphor; the Habit Engine runs the UPDATE INSTALL frame (SOS versions, patches, pending). They are compatible (a dose and an update are both "take it once daily") but pick which one leads each surface: the Today rail leads with MOMENTS/doses, the patch-note line and SOS version stay as the install layer's flavor. Do not let both metaphors fight in one sentence of copy.

10. BUILD MAP (who does what; wave-aligned)

v0 / MONDAY (in priority order)

WAVE 2 (weeks 1-3 of beta)

Overnight-delivery reveal flow + notify-me (email interim, push later), automatic status-flag wiring, Blueprint charts endpoint + screen (SVG reuse), written blueprint sections from the Understanding, Voice Lab (auditions + Spanish shelf), mic-next-to-send (Whisper interim), Oracle starters if not already in, A/B distributor + "try version B" cards, feedback-log "what changed" loop, /render-patch engine spec, Limiting Belief box, second + third constellations (Taurus or context-picked), request desk expansion, birth-time rectifier flow.

WAVE 3 (post-beta)

Relationships section + viral invite loop (spec 4e), Penta view, system constellations (GK/HD/Vedic), shareable badges + snippets, Mindset Ninja practice engine, own-voice experiment, dance/music tracks, hard-mode constellations, Spanish full experience, creator/Affirmologist dials.

11. SYSTEM STATUS + THE DELIVERY LOOP (Jeff's 7/11 addendum)

11.1 Honest system-status messaging (v0-cheap, do it Monday if possible)

Beta users must never stare at a spinner wondering if we broke. One reusable STATUS BANNER component (app-wide, driven by a tiny GET /api/app/status flag the team or the backend can set): - CREDIT/OUTAGE state: "A note from the team: our audio engine hit a temporary limit on the backend. Your audio is queued and WILL be created. This is beta; kinks like this get worked out. Check back in a bit, or we'll let you know." Warm, honest, on-brand, zero fake progress. - DEGRADED state (slow renders): "Renders are running slower than usual right now; yours is in line." - The flag can be set manually (an admin endpoint) at first; automatic detection (the credit-error detector already exists in self-heal) wires in later. The graceful-failure work on the results page (7/4) is the same philosophy; this brings it in-app. - Per-request status: every requested audio has a visible state in Get Audios: QUEUED → CREATING → READY (→ or WAITING ON THE ENGINE during an outage).

11.2 The overnight-delivery loop (renders land while they sleep; the morning reveal)

Renders are per-person and can trail the journey by a day (Map v2), so make the LANDING an event: - Overnight ingests (via /api/library/ingest) mark audios NEW; the app badges Listen ("1 update pending", the native Zeigarnik loop) and the Today rail shows the new card with its patch notes. - "NOTIFY ME WHEN IT'S READY": every request card carries it. Until push infra exists, this = the daily email carries "your new audio landed" (email is already the interim cue carrier per Map v2); when push notifications come online (Wave 2-3), it becomes a real push, and later the anchor-word push rides the same rail. - The reveal motivates exploring: a landed requested audio deep-links into Explore > Get Audios with a gentle "want another? here's what else can be built for you" row. Request → wait → reveal → request again is itself a habit loop, and an ethical one (real value each cycle).

11.3 Credits + account types (parked, but shape the data now)

Later there are account tiers (member / creator / etc.) and audio credits (included + purchasable) governing request volume. Do NOT build the economy now; DO log every request/render against the person with a cost estimate (the engine already tracks render cost), so when credits arrive the ledger and habits already exist. Beta = generous visible allowance (7.4), which doubles as the credits UI dress rehearsal.

12. ATLAS: THE CREATOR LAB (notes for the ~next-month build; NOT this weekend)

Jeff's direction, captured so it is not lost: - PROBLEM: Jeff currently gets more creation utility from Claude Code sessions than from Atlas. Atlas must become the attuned lab: create/tweak audio structures conversationally through Hermes AND through direct dials, no Code session required. - DIALS: the full per-audio metrics set from North Star section 12 (identity strength, transformation strength, energy/intensity, pacing, emotional tone, techniques employed, stage, time-of-day, theme, duration, modality lean, voice, bed, repetition role) as adjustable controls on every draft, adjustable directly or by telling Hermes. - THE TECHNIQUE BROWSER: a browsable, clickable LIST of every transformational tool/technique in the corpus (the 339+ craft entries + the S-index structures), each with a paragraph of what it is and when it serves; click to include it in a build. New corpus findings flow in automatically (the corpus IS the memory; Atlas becomes its visible shelf). Favorites, test combinations, retest, dial in, get suggestions from Hermes. - PIPELINE: draft in Atlas → test render → listen → tweak → PUSH to (a) your own app library, (b) a named person's library, or later (c) a community/creator audience. Same engine, same QC gate, always. - CREATOR ACCOUNTS (the tier from 11.3): graduated complexity. The consumer app stays streamlined with decisions made for you; power users upgrade to creator level and get Atlas access to build for themselves and their communities (the Affirmologist path). App = simple; Atlas = the lab. - This becomes its own spec doc when the build starts (working name: Affirmology_AtlasCreatorLab_Spec_v1.md); the existing Atlas decisions (PROJECT_STATE 17b, permission tiers, voicenote transcript-only) still stand.

13. OPEN DECISIONS FOR JEFF (small, answer whenever)

1. Explore v0 live-row set: agree with the 5 proposed (Constellations, Get Audios, Feedback Hub, Profile, Help)?
2. The "generic ↔ deeply me" slider as tap 2 at audio end: in for Monday?
3. Aries reward = guide activation (per the Gamification doc): confirm that is still the call.
4. RESOLVED by the AppProduct Principles daily architecture: FOUR daily doses (morning, meditation, movement, before sleep) + the reading as a WEEKLY card, not a fifth dose.
5. Spanish voice shelf: greenlight auditioning 2-3 Fish Spanish voices this week?
6. THE NAMING PROBLEM (from Sol, flagged blocking in the principles doc, needs a decision not drift): (a) does "Soul Song" survive (it is spoken, not sung; Sol and Mio both tripped); (b) one word per thing across the muddy set (soul song / reports / readings / tracks / audios / journeys). In-app, the moment-based sort (four doses + readings) hides most of the confusion, but the customer-facing names still need the call. Proposal to react to: free first audio = keep its proper name (Origin), ongoing daily audios = "your doses" informally and "audios" plainly, spoken self-knowledge pieces = "readings", the written PDF = "your Blueprint", chat = "the Oracle" (already clean).

Housekeeping: this doc is the app-experience lead; PROJECT_STATE points here. Update section 10's wave tables as items ship. The S-index stays in the Possibilities Maps; journeys catalog (4c) may graduate to its own doc when journeys build starts.