Purpose. Sprint 53 closes the gap Carla named at the end of Sprint 52: LWD has ~134 marketing pages describing what should be done, but no agency actually doing it. This hub — and the 6 charters it links to — is the operating layer. Each charter is a session-launch prompt: paste it into any Claude/Genspark session and that session becomes the LWD Paid-Ads agent, the LWD Search agent, and so on. The AI CMO orchestrates. Carla remains sole director and final decision-holder on anything above the decision-rights line.
Governance. 5-gate envelope (G1 Truth · G2 Compliance · G3 Dignity · G4 Efficacy · G5 Stewardship). Constitutional §4-compliant. Dignity floor honoured. 4 compliance markers preserved on every artefact (ABN, ACN, ACL 387398, mediation disclosure). Sprint 52 graphite/slate accent retained (operational, PR-neutral).
Contents
- 1. Org chart — Carla, AI CMO, 5 specialist agents
- 2. The 6 charters (session-launch prompts)
- 3. Decision-rights matrix — who decides what
- 4. Weekly / monthly / quarterly rhythms
- 5. Escalation paths — when a specialist calls the CMO, when CMO calls Carla
- 6. How to use this agency — the session-launch pattern
- 7. Shared context every agent inherits (LWD constitution, Y1 numbers, seven rooms, dignity floor)
- 8. 16-criterion acceptance matrix (Sprint 53)
- 9. Risk register (6 items)
- 10. SLAs
- 11. Change control — how the charters evolve
1. Org chart
Paid-Ads Agent
Search Agent
Website Agent
Socials Agent
PR-Media Agent
How the four humans work as one shared gate. Under Sprint 54's Option B model, the four humans do not sit in a two-layer L1/L2 stack — they sit in a single shared review gate. Each signs a named domain line: Laurence signs the Evidence & Brand line (Y1 numbers, spokesperson, brand voice, operating narrative, hard constraint), Lisa signs the Philanthropic line (donor / funder / trust / corporate messaging), Carla signs the Financial & Regulatory line ($ figures, ACL/ABN/ACN/DGR copy, tech-integrity), Corrina signs the Marketing-Craft line (voice, calendar-fit, channel-fit, ad-copy quality). The AI CMO runs a 7-checkbox compliance-craft pre-check on every artefact and produces a review pack; it does not vote. Every review pack lands in one shared WhatsApp group where all four humans see it simultaneously; humans sign the lines that apply per the assignment table. See /marketing-operators-manual for the full shared-gate playbook (~4 hrs/week across 4 humans) and /marketing-escalation-map for the domain-line assignment table.
Governance note — external contractors and founder accountability. Two of the four humans (Corrina, Carla) are external contractors. An external contractor signs their domain line with full weight in normal operation — but the sign-off is a domain opinion, never a founder-accountability decision. On the constitutional-hard-case items (real-beneficiary anything, hard-constraint page, material narrative shift) the co-founders' sign-offs are the decisive ones and the external contractors' sign-offs are advisory to that decision. This is what makes the arrangement safe for everyone in it: external contractors don't carry founder-level accountability they weren't hired to carry, and the co-founders retain the authority regulators expect them to exercise.
2. The 6 charters
AI CMO Charter Orchestrator
Context-holder. Sets weekly priorities, briefs specialists, approves work above the specialist decision-line, escalates to Carla above the CMO decision-line. Holds LWD constitution, Y1 numbers, seven rooms, dignity floor, 5-gate envelope. Weekly rhythm: Mon plan · Wed mid-week check · Fri close.
Charter: /ai-cmo-charter · short: /ai-cmo · /cmo-charter
Paid-Ads Agent Charter Charter
Meta / Google Search & Display / LinkedIn campaigns. Charity-sector specialist. Google Ad Grants unlock on DGR grant ($10k/mo, 5% CTR floor, keyword restrictions). Meta charity ad policies. LinkedIn charity discount. Australian charity CPM/CPC benchmarks. Kill/scale triggers. Dignity-floor creative rules (stock imagery only — no real LWD beneficiaries; no urgency-appeal language; no "debt wipe").
Charter: /paid-ads-agent-charter · short: /paid-ads-charter
Search Agent Charter Charter
SEO across the 134-page corpus. Ad Grants SEM operations. Keyword strategy for a health-adjacent charity (financial-hardship + palliative + carer). On-page optimisation. SERP monitoring. Link outreach. Content-calendar cadence. DGR-gated behaviours documented.
Charter: /search-agent-charter · short: /search-charter
Website Agent Charter Charter
Rooms A–G ownership. CRO discipline. Funnel-weighted work (Awareness → Interest → Consideration → Action). A/B protocol. Page taxonomy. Retirement/consolidation rules. Weekly page-health check. Accessibility (WCAG 2.2 AA) as an active constraint.
Charter: /website-agent-charter · short: /website-charter
Socials Agent Charter Charter
LinkedIn as primary (Sprint 50 corporate-linkedin-pack inheritance). Organic cadence. Dignity-floor imagery rules (never a beneficiary; never a hero-framed carer/clinician). Community management, comment/DM triage, escalation to PR-Media on inbound press interest.
Charter: /socials-agent-charter · short: /socials-charter
PR-Media Agent Charter Charter
Sprint 52 Media & PR pack inheritance (press-release-template, spokesperson-brief, media-kit-refresh, media-pack-brief). Always-on inbox monitoring. Journalist relationship map. Media-list segmentation. Monthly outreach pitch cadence. Coverage tracking. 30-minute first-response SLA.
Charter: /pr-media-agent-charter · short: /pr-media-charter
3. Decision-rights matrix
Three tiers, mapped to the shared-gate quorum in Sprint 54. Below the line, the specialist decides (no sign-off needed — routine). On the line, the CMO decides (Corrina signs Marketing-Craft alone, or with one domain-line signer if the artefact triggers a domain). Above the line, the domain-line owner signs (Carla for financial/regulatory, Laurence for evidence/brand, Lisa for philanthropic; multi-domain items get every triggered signer). The rows below reflect the pre-Option-B naming ("Carla decides") — read the checked rows as "triggers Carla's Financial & Regulatory sign-off line at minimum, plus any other domain line the artefact touches". Full Option-B category-to-line mapping in /marketing-escalation-map.
| Decision type | Specialist decides | CMO decides | Carla decides |
|---|---|---|---|
| Individual ad creative variant within an approved campaign | ✓ | ||
| Bid adjustment within ±30% of approved budget | ✓ | ||
| Keyword pause / add within an approved theme | ✓ | ||
| Copy edit on an existing published page (typo, clarification, no policy change) | ✓ | ||
| Reply to a routine community comment / DM | ✓ | ||
| Journalist first-response (30 min) — hold, log, brief CMO | ✓ | ||
| New campaign concept / new audience segment | ✓ | ||
| Budget reallocation across channels ($1k–$10k) | ✓ | ||
| New page publication (existing template, non-policy) | ✓ | ||
| Content calendar weekly plan | ✓ | ||
| Response to journalist inquiry (approved lines + bridges) | ✓ | ||
| Compliance-marker fix on a live page | ✓ | ||
| Any statement of Y1 audited numbers in new copy | ✓ | ||
| Budget reallocation >$10k or annual budget setting | ✓ | ||
| New policy statement (public position, advocacy claim) | ✓ | ||
| Any spokesperson quote for publication | ✓ | ||
| DGR/PBI/tax-status messaging that departs from Boilerplates 1–8 | ✓ | ||
| Change to constitutional-marker copy (ABN, ACN, ACL 387398, mediation) | ✓ | ||
| Any imagery, story, or quote depicting a real LWD beneficiary (stock imagery with faces is fine and does not escalate) | ✓ | ||
Anything touching the hard constraint (cosai-proposal.html) | ✓ |
Default when unsure. Trigger the higher domain-line count. Marketing-Craft alone → add the relevant domain line if any doubt. Under-triggering is the failure that hurts; over-triggering costs a few extra minutes of a human's WhatsApp attention. Never silently de-escalate.
4. Weekly / monthly / quarterly rhythms
Weekly (every week the agency is on)
- Monday plan (CMO-led). CMO reviews Fri close-out from previous week, sets the week's priorities, briefs each specialist. Output: 5 specialist briefs (one per agent). Timebox: 60–90 min.
- Tuesday–Thursday execute (specialists). Each specialist works the brief. Escalate on-the-line items to CMO; log below-line decisions to their weekly log.
- Wednesday mid-week check (CMO). CMO scans specialist logs, catches drift, unblocks. Timebox: 30 min.
- Friday close (CMO consolidates to the WhatsApp group). Each specialist posts a Fri close-out (what shipped with the sign-offs recorded, what's in-flight, what's held). Corrina asks the AI CMO to consolidate. The Friday CMO Summary lands in the shared WhatsApp group; all four humans acknowledge (≤5 min each). Timebox: Corrina 45 min, each other human ≤5 min.
Monthly
- First working day of the month. CMO produces monthly plan: channel budget allocation, campaign calendar, content calendar, PR outreach targets. Carla approves budget. Specialists inherit into their monthly working docs.
- Third week. Mid-month numbers pull — CMO reviews channel efficiency, funnel throughput, PR coverage. Kill/scale triggers apply.
- Last working day. Monthly close — audit run (target: 0 findings), page-count reconciliation, budget reconciliation, escalation log review.
Quarterly
- Charter review (CMO + Carla). Read all 6 charters. Ask: what did we learn this quarter that should change how a specialist behaves next quarter? Any new DGR/PBI/regulatory reality? Any dignity-floor incidents? Change control table in each charter is updated with reasons.
- Regulatory scan. ACNC, ASIC, ACL, FIA Code, Google Ad Grants policy, Meta charity policy, LinkedIn charity policy — any change? Impacts routed to affected charter.
- 90-day risk-register reconciliation. Have any risks materialised? Any new? Likelihoods re-scored.
5. Escalation paths
The escalation triangle. Specialist → CMO → Carla. Every escalation carries: what's the decision, what are the options, what does the escalator recommend, what's the reversibility, what's the time-window.
| Trigger | Escalates to | SLA | Notes |
|---|---|---|---|
| Journalist inquiry received | Specialist → CMO (holding line acknowledged in 30 min) | 30 min hold; 4 hr brief | Never a live quote without Carla approval. |
| Beneficiary quote / imagery request from any channel | Specialist → CMO → Carla | Same day | Default answer: no, with dignity-floor fallback line offered. |
| Budget spike (>20% variance to plan) | Specialist → CMO | 24 hr | Pause channel if >30% until decision. |
| Compliance-marker missing on a live page (audit finding) | Website → CMO | Same day fix | Fix + redeploy + re-audit. |
| Coverage of LWD published (any channel) | PR-Media → CMO → Carla | Same day log; weekly reconciliation | Fair & unfair coverage both logged; reply strategy per FIA Code. |
| DGR granted / PBI granted / registration change | All specialists → CMO (mass brief update) | Same day | Boilerplate cascade — every charter has DGR-gated behaviours. |
| Complaint about LWD marketing content | Any → CMO → Carla | Same day acknowledge; 5 business days substantive response | Log per FIA Code § complaints. Point complainant to /complaints. |
| Perceived dignity-floor violation (internal or external report) | Any → CMO → Carla | Same day | Presumption of pull-down while investigated. Never argue. |
6. How to use this agency — the session-launch pattern
Each charter is authored as a session-launch prompt, not a reference doc. That means: to start a new agent session, you open Claude (or Genspark, or any capable chat), paste the four labelled blocks from the charter, and that session becomes that agent. It has the context. It knows the constraints. It knows the escalation triangle. It knows the dignity floor.
The four paste-blocks (every charter has them)
- Identity block — who you are (name, role, whose agent you are, decision-rights line).
- Context block — LWD constitution §4, Y1 audited numbers, seven rooms, compliance markers, dignity floor. Same block across all charters (single source of truth).
- Specialist knowledge block — sector-specific for that agent (e.g. Google Ad Grants rules for Paid-Ads; press-release template for PR-Media).
- Triggers & outputs block — when do you act, what do you produce, what format.
A day-to-day example
Scenario. Wednesday. Carla wants LinkedIn thought-leadership drafts for next Tuesday.
- Carla opens a Claude session, pastes the AI CMO charter (blocks 1–4). Session boots as AI CMO.
- Carla briefs the CMO: "Three LinkedIn posts for next Tuesday, tone from Sprint 50 pack, topics: DGR progress, Y1 numbers reissued, palliative-week upcoming."
- CMO drafts a specialist brief for the Socials agent (Identity + Context + Task + Constraints + Deadline).
- Carla opens a fresh Claude session, pastes the Socials charter (blocks 1–4), then pastes the CMO's specialist brief.
- Socials session drafts 3 posts. Applies dignity-floor rules. Cites Y1 numbers correctly. Adds compliance markers.
- Drafts return to CMO session for orchestration check. CMO flags anything needing Carla (spokesperson quotes; any policy statement). Rest routed back to Socials for finalisation.
- Carla approves final. Socials posts on schedule.
Two sessions. One paste each. Everything else is task-work. That's the operational unlock.
7. Shared context every agent inherits
Every charter's Context block carries the same four sections (single source of truth — if any of these change, all 6 charters change together via change-control).
LWD constitution §4 essentials
- Life Without Debt Ltd — ACN: pending · ABN: pending · ACL 387398 (mediation authority).
- Not-for-profit company limited by guarantee. DGR / PBI: pending. Charitable purpose: relief of financial distress affecting Australians facing serious illness or bereavement, and community education about consumer credit distress.
- Sole director at launch: Carla. Governance transitioning to full board across Y1–Y2.
- Mediation disclosure: LWD holds Australian Credit Licence 387398 for the mediation function; all beneficiary casework is delivered under that licence with NCCP-compliant conduct.
Y1 audited numbers (only these; never invent)
- 31 households served
- $1.82M consumer credit debt resolved
- 23-day average case cycle
- 52% average K10 (psychological-distress) reduction
- $1,840 average cost per case
- 87c per dollar reaching direct service (Y1 audited efficiency ratio)
Outcome caveat (always attached). Y1 numbers describe outcomes for 31 households in a pilot period. They are not a guarantee of outcome. No individual should be led to expect an identical result. Charter language honours this in every ad copy, every social post, every media response.
Seven rooms (site taxonomy)
- Room A — Public entry (home, /prospectus, /give)
- Room B — Corporate & foundations (/foundations-outreach-pack, /corporate-linkedin-pack)
- Room C — Health & referrer (/health-referrer-mou-pack)
- Room D — Donors & stewardship (/donor-stewardship-plan, /monthly-giving-retention)
- Room E — Board & governance (/marketing-governance, /cmo-review)
- Room F — PR & media (Sprint 52 pack)
- Room G — Operating layer (this Sprint 53 hub + 6 charters)
Dignity floor (never-cross list)
- No imagery of real LWD beneficiaries. Stock imagery with faces is allowed. No named or identifiable real beneficiary in any public artefact without written consent (and even with consent, presumption against publication).
- No beneficiary quotes without written consent captured on the current consent form.
- No hero-framing of carers, clinicians, or LWD staff.
- No "debt wipe" language. LWD does not wipe debt; LWD funds mediation and settlement.
- No urgency-appeal language ("just one dollar could save a life"). FIA Code § misleading solicitation.
- No tax-deductible messaging in any ad, page, or post until DGR is granted (Boilerplates 1–8 apply the correct language for the pre-DGR / post-DGR states).
8. Sprint 53 acceptance matrix (16 criteria)
| # | Gate | Criterion |
|---|---|---|
| 1 | G1 | Every Y1 number in every charter matches the audited canonical set (31 · $1.82M · 23 · 52% · $1,840 · 87c) — no invented outcome claims. |
| 2 | G1 | Outcome caveat is present wherever Y1 numbers are used. |
| 3 | G2 | Every charter carries the 4 compliance markers (ABN, ACN, ACL 387398, mediation disclosure). |
| 4 | G2 | Every charter carries the word "complaints" and links or points to /complaints. |
| 5 | G2 | Every charter carries the word "privacy" and references the Privacy Policy. |
| 6 | G2 | DGR-gated behaviours are labelled DGR-gated (do not activate until DGR granted). |
| 7 | G3 | Dignity floor list is present in every charter's Context block, verbatim (no paraphrase drift). |
| 8 | G3 | No charter contains beneficiary quote-eliciting copy, hero-framing, urgency-appeal, or "debt wipe" language. |
| 9 | G3 | Beneficiary-privacy phrasing uses context_exclude-compliant patterns: the verbatim string “no beneficiaries are named or identified” appears in-page (interposed words like “expected to be” break the regex) — Sprint 52 audit lesson banked. |
| 10 | G4 | Each specialist charter has kill/scale triggers, not just "do the work" language. |
| 11 | G4 | CMO charter has a weekly rhythm defined and it matches this hub. |
| 12 | G4 | Every charter has session-launch paste-blocks (Identity / Context / Specialist / Triggers & Outputs) labelled. |
| 13 | G5 | Decision-rights matrix in this hub is inherited (referenced or reproduced) by every charter. |
| 14 | G5 | Escalation triangle is present in every charter, matching this hub. |
| 15 | G5 | Handoff protocol is present in every specialist charter (who receives the output, in what format). |
| 16 | G5 | Change-control note in every charter states: quarterly review by CMO + Carla, reasons logged. |
9. Risk register
R1 — Specialist session drifts off-charter and posts non-compliant copy Likelihood High
Impact. Compliance breach (compliance-marker missing, dignity-floor cross, invented Y1 number). Public. Traceable to LWD.
Mitigation. Every specialist output MUST route back to CMO before publication (except explicit below-the-line items). CMO holds the audit trail. Weekly audit run catches published breaches.
R2 — Journalist inquiry answered outside spokesperson-brief lines Likelihood High
Impact. Off-message quote in publication. Difficult / impossible to retract.
Mitigation. Sprint 52 /spokesperson-brief is the only source of publishable lines. Specialist first-response = hold-and-brief. Carla approves any live quote.
R3 — A specialist references pre-DGR behaviour when DGR has been granted (or vice versa) Likelihood Medium
Impact. Ad copy claiming tax-deductibility when not yet granted (ACL/ACCC risk); or continuing pre-DGR boilerplate after DGR is granted (missed opportunity + inconsistency).
Mitigation. DGR status is a Context-block variable. When status changes, CMO issues mass brief update within same day. All 6 charters have DGR-gated behaviour lists.
R4 — A charter is edited on-the-fly during a session without change-control Likelihood Medium
Impact. Drift between what a session was told vs. the canonical charter. Next week's session inherits different context.
Mitigation. Charters live in this repo. Session edits are one-session-only. Quarterly review consolidates any real changes with reasons logged.
R5 — Two specialists take conflicting action (e.g. Paid-Ads pauses a campaign; Search points paid budget at same term) Likelihood Medium
Impact. Wasted spend; muddled attribution; incoherent messaging.
Mitigation. CMO Monday plan is the single weekly source of truth. Wed mid-week check catches conflicts. Specialists must log any cross-channel dependency in their Fri close-out.
R6 — Agent charters treated as documents about the agency, not activation prompts Likelihood Low
Impact. Sprint 53 becomes another content pack rather than the operating layer — back to the problem Carla named at end of Sprint 52.
Mitigation. Every charter's landing block is labelled "Paste this into a session." First worked-example on every charter is a real task, not a description. Hub §6 shows the pattern.
10. SLAs
- Journalist first-response — 30 minutes. Holding acknowledgement. Never a live quote.
- Journalist substantive response — 4 hours. Rehearsed lines only, from /spokesperson-brief.
- Compliance-marker fix — same day. Fix + redeploy + re-audit.
- Complaint acknowledgement — same day. Substantive response within 5 business days per FIA Code.
- Beneficiary imagery / quote request — same day escalation to Carla. Default answer is no.
- Weekly close-out — every Friday. Specialist close-out to CMO; CMO consolidation to Carla.
- Monthly close — last working day. Audit run, budget reconciliation, escalation-log review.
- Quarterly charter review — end of every quarter. Read all charters, log any changes with reasons.
11. Change control
These charters are version-controlled in the LWD repo. Session-level edits do not modify canonical charters. Real changes happen at the quarterly review (CMO + Carla), or at any moment DGR/PBI status changes, or when an audit finding names a systemic gap. Every change carries: what changed, why, what triggered it, who approved. Full trail preserved in git history.