Who this is for. Lisa (Co-founder & Head of Philanthropy) is the answerer. Corrina is the courier — she reads this brief into the WhatsApp group, waits for Lisa's answers, then feeds them to the AI CMO who folds them into the seven donor pages via the domain-line assignment table. Laurence and Carla see the brief in-thread so they know what's coming.
Why this is the Sprint 55 shape. Room D already has seven artefacts on disk (~120 kB of donor copy: /donor-one-pager, /donor-stewardship-plan, /foundations-grant-briefing, /foundations-outreach-pack, /founding-donor-deck, /give, /give-sprint48-brief). Structure is there. Voice is thin. Sprint 55 is not "build room D from scratch" — it is "thread Lisa's actual voice through the existing seven pages." That is a surgical operation: 5–7 paragraphs change, not 120 kB of copy.
Companion pages. /donor-voice-map shows exactly which paragraph in which page each of the seven answers lands in — the AI CMO uses that map as the substitution index. /edit-sheet is the read-and-correct alternative to answering cold: AI-drafted answers on the left, Lisa's inline corrections on the right, one row per question, print-friendly for A4 markup. The draft is a scaffold Lisa reacts to, not a substitute for her voice — nothing ships to donor pages until Lisa signs the Philanthropic line.
Compliance envelope. Y1 audited numbers verbatim; dignity floor intact; no beneficiaries are named or identified in Lisa's answers or in the resulting donor copy; ABN, ACN, ACL 387398, mediation disclosure carried on every downstream page.
Contents
1. How to answer — the mechanics
The easiest path is WhatsApp voice-notes. Seven questions, one voice-note per question, each 60–120 seconds. Total ~15 minutes if answered in one sitting. Voice-notes preserve Lisa's cadence and hesitations, which is the point. Corrina transcribes; the AI CMO extracts the substantive sentences and folds them into the donor pages per the voice-map. Lisa reviews the resulting paragraphs before anything ships (the Philanthropic sign-off line).
Text answers are fine too. One WhatsApp text message per question. Bullet points are fine. Fragments are fine. Complete sentences aren't required — the AI CMO does the sentence assembly.
"I don't know yet" is a valid answer. Some questions may genuinely not have a Lisa answer today (e.g. Q7 depends on a real conversation with a specific foundation trustee that hasn't happened). Mark those "not yet" and we'll route around them until the answer exists.
2. The voice we are trying to capture
Not a scripted charity voice. Lisa's actual register, which across the constitutional docs and the co-founder conversations reads as: warm but non-appeal, evidence-led, qualified rather than certain, quiet rather than urgent, and speaks about a specific person even when the person is anonymous (never "Australians facing hardship" — always something closer to "a woman I sat with in June"). The seven questions are engineered to elicit that register. If an answer starts drifting into standard charity phrasing, the question was probably wrong — flag it and we'll rewrite.
The dignity floor still applies inside voice-capture. Even in Lisa's answers, no beneficiaries are named or identified. If a specific person is being described — and often they will be — the identifying details are stripped by Corrina before the text reaches the AI CMO. "A woman I sat with in June" is fine. A first name, a suburb, a diagnosis, or a debt figure paired with a household size is not.
3. The seven questions
Read the question, then answer. The context line explains why we're asking; skim it or skip it. The example register shows the kind of voice we're trying to capture (not a script — a tuning fork). The answers into line names which donor page(s) will absorb the answer.
Why does LWD exist for you personally?
60–120sThe founder story sentence. Every donor page opens with some variant of "why this organisation." Right now that opener speaks in charity-generic. We need Lisa's version — the specific reason she is doing this rather than something else.
What is the moment you'd want a foundation trustee to picture?
60–120sFoundation trustees read hundreds of grant briefings a year. Ours needs one anchoring image — something specific, dignified, non-appeal. The image belongs to Lisa (not a beneficiary), which means it can be public without a consent problem.
What does the mediation function actually do, in your words?
60–90s"LWD holds Australian Credit Licence 387398 for the mediation function" is the compliant sentence. It's also the sentence that shuts a donor's brain off. We need Lisa's translation — what mediation looks like from the perspective of the household on the other side of it.
Why 31 households and $1.82M — and not something bigger?
45–90sThe Y1 audited numbers are the truth we lead with. But a foundation trustee's next thought is "why so few?" and the answer is not "because we couldn't help more" — it is a deliberate choice about proving the model before scaling. That reasoning needs to be in Lisa's voice, not the charter's.
What is the wrong reason to give to LWD?
45–90sEvery charity page tells donors why to give. Almost none tell donors why not to give. That's what makes LWD's dignity floor visible — we're the org that says "if you want a story of a specific rescued family, we're not that org." That negative frame does more donor-trust work than a positive one.
What does year two look like if it works?
60–120sEvery donor page ends with a forward-look. Right now it reads as growth-rhetoric. It should read as Lisa describing what she expects to be true in twelve months if the DGR grant lands and the model scales the way the pilot suggests it will — qualified, not certain.
The direct answer to "why should I trust you with this?"
45–90sEvery foundation trustee, every founding donor, every corporate-CSR lead will at some point ask a version of this question. It's not a hostile question; it's a professional one. Lisa's answer to it is the single most important sentence in room D — and right now the site doesn't contain that sentence in her voice. This question closes that gap.
4. Anti-patterns — what to avoid saying
Not because Lisa would say these — but because if the questions elicit these, the questions were designed wrong and we need to rewrite them, not use them.
- Urgency-appeal language"Every day we delay, a family loses everything." — This is charity-generic and ASIC RG 96 discourages it. Lisa doesn't talk like this; the copy shouldn't either.
- "Debt wipe" or "debt relief" framingLWD funds mediation and settlement, not debt extinguishment. Using "wipe" would be misleading and RG 96-breaching. If an answer drifts here, Corrina translates before it reaches the AI CMO.
- Naming or identifying a specific beneficiaryEven one identifying detail (first name + suburb + illness) is a dignity-floor breach. If Lisa describes a specific person, Corrina strips the identifiers to render the description generic-anonymous ("a woman I sat with in June") before the text moves.
- Hero-framing carers, clinicians, or LWD staffThe dignity floor bans this in both directions. Lisa is the founder answering questions, not the hero of the story. The households are not the heroes either. The model is the hero. Keep the voice focused on what LWD does rather than who is admirable.
- Tax-deductible messaging pre-DGRUntil DGR is granted (imminent, but not today), no answer can imply tax-deductibility. If an answer touches "tax-deductible," Corrina flags it and we wait for DGR before folding that sentence into donor copy.
- New $ figures beyond the Y1 canonical setLisa may reference the Y1 numbers verbatim (31 · $1.82M · 23-day · 52% · $1,840 · 87c). Any new financial claim (year-two budget, per-household cost projection, staff cost) needs Carla's Financial & Regulatory sign-off before it lands in donor copy — it does not fold in automatically from a Lisa voice-note.
- Guarantees about future outcomes"We expect" and "we intend to" pass. "We will" and "we guarantee" trigger a Financial & Regulatory line for Carla. Q6 is engineered around this — if the answer drifts to certainty, Corrina softens it back to qualified language.
5. What happens after Lisa answers
The Sprint 54 shared-gate machinery already exists; Sprint 55 rides on it.
- Corrina couriers Lisa's seven answers from Lisa's WhatsApp voice-notes or texts into the shared HA5 WhatsApp group. Voice-notes are transcribed; identifiers (names, suburbs, diagnoses, specific debt figures paired with household size) are stripped in the transcription pass.
- The AI CMO extracts the substantive sentences from each answer and uses /donor-voice-map as the substitution index — each answer maps to specific paragraphs in specific donor pages.
- The AI CMO produces a review pack per donor page (6-item bundle, per the Sprint 54 pattern): the current paragraph, the proposed paragraph, the source Lisa answer, the domain lines triggered, the quorum level, and the reversibility note. Each page's pack is a separate review pack — not one giant one.
- The domain lines triggered are typically Philanthropic (Lisa signs her own voice back in — this matters, it's not a rubber-stamp; she reads the AI CMO's assembly of her voice and confirms) and Marketing-Craft (Corrina signs the cadence and channel fit). For paragraphs that touch Y1 numbers or forward-looking financial claims, Financial & Regulatory (Carla) also triggers. Constitutional if any paragraph accidentally touches the dignity floor.
- Quorum is domain-touching for most paragraphs (Lisa + one second-eye human), multi-domain where Y1 numbers or Y2 projections appear (Lisa + Carla + one second-eye), constitutional if the AI CMO's assembly accidentally produces identifying content (all 4 humans).
- Once quorum is complete, Corrina posts the finalised paragraph substitutions in the WhatsApp group and the seven donor pages are updated in a single Sprint 55 commit. Rebuild · verify hard-constraint sha256 · deploy · smoke-test.
6. Session-launch paste (for the AI CMO)
When Lisa's answers land in the WhatsApp group and Corrina is ready to feed them to the AI CMO, this is the paste that opens the session. It sits alongside the four standard Blocks from the AI CMO charter.
7. Change control
The seven questions above are Sprint 55's contract. If any question needs to be rewritten (because it isn't eliciting Lisa's voice, or because Lisa says "that question is wrong"), the rewrite is a Sprint 55 amendment and lands in the shared WhatsApp group as a review pack: Philanthropic line (Lisa signs the question she's being asked) + Marketing-Craft line (Corrina signs the question quality) + one second-eye. Anti-patterns (section 4) are governance items — adding or removing an anti-pattern triggers Constitutional quorum (all 4 humans sign). Rewriting the session-launch paste (section 6) is Marketing-Craft alone (Corrina signs; second-eye Laurence or Carla depending on which frameworks it touches).