Home / Ops / More in this area
Updated Jul 04, 2026 · Affirmology_FlagshipReading_Proposal_v1.md
Written 2026-07-03 by Cowork, answering Jeff's questions: (a) what would a REALLY landing 10-12 minute audio look like, (b) how do we imitate the Lakshmi-type reading cheaply to load epic context into ALL audios, (c) what did the recovered transcripts teach us, (d) what does this do for affirmational audios including the Origin Soul Song. Companion docs: Affirmology_ReadingCraft_HeroStory_Synthesis_v1.md (the craft standard), Affirmology_WealthReading_Transcript_Jeff_v1.md and _Sol_v1.md (the recovered exemplars), Affirmology_AhHaMoves_Registry_v1.md (the aha moves). No em dashes.
Beyond everything in the car-notes synthesis, the full transcripts show these repeatable moves. Verbatim evidence in parentheses:
Flaws the transcripts also revealed (QC additions): - Composer seams left in the audio. Lines like "I have everything I need from Sophia. Let me now write the full polished section two" and "Now I have everything I need. Let me weave..." were spoken aloud in the final render. The QC gate must strip any composer/meta narration. - Truncated ending. Jeff's 72-min MP3 ends mid-sentence at 71:37. Add an end-integrity check (final sentence completeness) to audio QC.
Yes to Jeff's instinct: the spine is CROSS-SYSTEM CONVERGENCE, and the flesh is the full technique stack above. The design:
Working name: The Mirror v4 (or "The Recognition Reading"). ~1,500-1,700 words at Storyteller pace.
Minute map: - 0:00-1:15, The Threshold. The moved reader opens ("what I am holding here is not a simple chart"). The structural-fact anchor (their sky, date, place). The hero frame: one paragraph of their life as tension and becoming. Then THE PROMISE, the Big 3 phrased as lived experience with curiosity gaps ("three things in this chart will not stop talking about you; the third one, I suspect nobody has ever said to you out loud"). Progress scaffold: "three doors, we will walk through all of them." - 1:15-4:15, Door One (the loudest convergence). Full segment template: felt pattern first, multi-domain examples, then the reveal with the convergence refrain ("four systems, built centuries apart, same address"), the mechanism, one extended bespoke metaphor, low road / high road contrast, the portable law, the practical move, the affirmation line spoken twice. Somatic checkpoint at the deepest beat. Rehook bridge out: promise Door Two as a paradox. - 4:15-7:15, Door Two (the juiciest tension). Same template. This is the shadow-to-gift door: fear preemption move, the reframe (aha move #10), the polarity conundrum ("how the thing you have apologized for becomes the thing they pay you for"). Rehook out: "and the third door explains why the first two have been fighting." - 7:15-10:15, Door Three (the emerging promise). Same template, aimed at North Node / gift-in-shadow / the becoming. The gap named, the path shown, timing touched if a real transit is live ("this is not a someday chart; here is what is turning right now"). Crescendo passage here, the passion peak. - 10:15-11:30, The Seal. The three affirmation lines restated as one breath-paced litany. The hero's summary sentence ("this is the story your sky has been telling since [birth date]"). One deliberate open loop pointed at the next audio ("there is a fourth thing in your chart. It is quieter than these three, and it is my favorite. Next time."). No CTA.
Hard rules carried in: never the checklist walk, never naked jargon before lived description, jargon demotion when a term is used, minimum two rehooks, one somatic checkpoint, one extended metaphor per door, every door ends in law + move + affirmation, composer-seam strip, end-integrity check, slightly slower pace with real pauses.
The insight from what actually happened: the wealth audios were great because they did NOT generate from the raw profile. The pipeline was: agent council (Lakshmi, Agastya, focused 1-2 oracle calls) -> a compiled context document (the WealthCodes docs) -> seeded into the knowledge base -> longform composition drew on that. The Mirror failed because it skipped the middle: raw chart JSON -> one flat pass -> generic garbage. The basic profile must be A MAP EVERYTHING ELSE USES TO GO GET BETTER CONTEXT, never the sole source.
DECIDED (Jeff, 2026-07-04): this ships, named THE UNDERSTANDING, and it is built INSIDE the Mirror/reading-craft rebuild, not as a separate project. Code's framing, which Jeff approved: the chart capture is solid; the one real tweak is that the deep read must stop being a single prose blob and become a structured, growing Understanding. The rebuild is already defining what a great reading needs (salient findings, convergences, aha moves), which is exactly what the Understanding stores, so building them together yields the cheap-reuse architecture for free. The council's job is to build a stored Understanding of the person ONCE, not to write audios. Every audio after (Mirror, Origin, bedtime, wealth, custom, chatbot answers) is a cheap shaping: select facets (context) + apply presentation (craft/journey). When we learn big things, we APPEND to the Understanding, never recompute. Two tiers stand: Tier 1 light and cheap (everyone gets the Origin), Tier 2 the deep council Understanding when someone commits, which then makes everything they ever receive cheap and epic.
The Understanding, minimum schema (v1):
1. salient_findings (ranked; Loud/Activated/Juicy scores per anchor; the Big 3 flagged)
2. convergences (cross-system agreements: shared truth + witnesses per system + felt-pattern sentence)
3. aha_moves (which registry moves this chart best supports, with the specific material for each)
4. wound_gift_gap (the core wound, the gift inside it, the named gap to who they are becoming)
5. mechanisms (causal chains in plain language)
6. roads (low road / high road pairs per major anchor)
7. metaphor_bank (bespoke images from this chart, never shared across people)
8. one_liners (affirmation-grade lines)
9. fear_preemptions (scary-sounding placements + honest reframes)
10. timing (transit/dasha windows with dates; refreshed monthly/new-moon)
11. context_store (intake, chat fragments, voice notes; grows over time)
12. learned_info (what landed/flopped per audio, structure feedback, quotable-back lines; grows over time)
The proposal detail as originally drafted (one Deep Map per person, built once, used by everything):
Cost picture: the Deep Run is a handful of focused agent calls once per person (the same shape as the five WealthCodes readings, which we know the funnel handles). Everything downstream drops to single cheap passes. This is strictly cheaper than today's per-audio improvisation AND strictly better, because every audio inherits the same epic, verified, chart-grounded context.
Tiering (who gets a Deep Run, Jeff's question 2026-07-03): do NOT deep-run everyone, and do NOT rely on the basic profile alone for anything paid. Three tiers: - Demo / funnel people: no Deep Run. The Origin Soul Song and teaser assets live on Enduring identity (Sun, Rising, Moon, Life Path, Incarnation Cross), which the basic profile already carries. One light salience pass (a single cheap call that scores Loud anchors and drafts the felt-pattern sentences) is enough to keep the demo non-generic. Cost stays near zero at the top of the funnel. - Committed beta testers and subscribers: full Deep Run on commitment. The moment someone opts in, the Deep Run fires overnight and their Deep Map exists before their first real reading. This is the basis for every epic reading and audio they receive, and it is exactly the investment that makes them stay. First-week retention is the whole game, so this is the right place to spend. - On-demand lenses: mini deep runs, cached forever. When a subscriber asks about something their map does not cover yet (a breakup, a launch, a health season), run ONE focused council call for that lens, append it to the Deep Map, and it serves every future audio too. Pay once per person per lens, never per audio.
The rule that keeps slop out: no paid audio ever composes from the basic profile alone. The profile routes; the Deep Map feeds. If the map lacks the needed lens, the mini run happens first, and if the council is unreachable the audio waits (per the standing hard rule, no improvising).
The same craft transfers to first-person Soul Songs, compressed:
Origin Soul Song gold script audit (Jeff's approved v1, ~430 words):
What already works: identity-first open, name by sentence 2-3, "I am the one who," the somatic stone-in-the-lake image, the hand-on-heart breath, gratitude beat, open-loop close. Keep all of it.
Where it shows the Mirror disease in miniature: 1. It tours placements. Virgo Sun/Venus 8th, then Jupiter/Mars in Scorpio 10th, then Life Path 22, each getting 2-3 lines. That is coverage, not thesis. FIX: pick the single convergence (for this chart, the Gatherer/Master Builder thread) and let every placement appear only as a witness to it. Say the convergence out loud: "every map I was written in says the same thing." 2. Its strongest image is under-used. The stone finding the lake floor appears once and is abandoned. FIX: one extended metaphor that opens, deepens, and closes the audio. 3. A few lines drift generic. "Structures that outlast trends," "something larger than any one person could hold" could be anyone. FIX: swap for mechanism lines and structural links, which are specific by construction. 4. No contrast beat. Nothing says what others do vs what I do. FIX: add one. 5. Bug: the script ends "I am already becoming her" in JEFF's approved copy. Wrong pronoun in the gold reference. Fix the reference file and add a pronoun check to script QC.
These fixes fold naturally into The Recognition A/B (the structure built exactly for recognition-density testing). Recommendation: apply this audit to The Recognition variant first, A/B against current Origin, let the test decide.
CLAUDE OUTPUTS/Affirmology/Affirmology_OriginSoulSong_Jeff_ApprovedScript_v1.md.