REFUTER — Deliverability & Credibility Prosecution

DPSC GO8264 six-application portfolio (DAS)

Stance: sceptical Commonwealth assessor + sceptical board member. Six submissions are LOCKED — this document optimises within six, it does not relitigate the count. Written: 11 Jun 2026. Deadline: 2pm AEST 2 Jul 2026 (15 working days). DASP tender (NDAP successor — existential) closes 16 Jul. Inputs: all six concepts (stream1/2/3-concepts.md), past-apps-digest.md, landscape-scan.md. Note on figures: market-rate estimates below (travel unit costs, interpreter rates, salary benchmarks) are prosecution estimates for re-costing direction — verify against current AIS/SCHADS/quote data before changing any budget line.

0. Executive verdicts

Portfolio posture (summary): submit all six, tiered — three full-dress (S1A, S2A, S3A), two descoped (S2B, S1B), one minimum-viable (S3B) — with the cross-app coherence devices in §6 and a board-resolved acceptance hierarchy. Detail in §6.

1. Portfolio math — what the assessor and the board member each see

The raw numbers

DAS today: ~7 staff, ~$1M revenue, net assets ~$197k (FY23 as last submitted), one GM, no operations manager, no grants-administration function, finance outsourced (Breakthrough Office).

The assessor’s view

The form requires disclosing the number of applications. Every assessor sees “6”. Six is the legal maximum (2 per stream: 1 individual + 1 consortium lead) — so the count itself is defensible as “we used the structure the department designed”. What is NOT defensible is six applications that each silently assume exclusive use of the same finite inputs:

• the same ~30-person Leadership Group as the §5.1 involvement mechanism in all six;
• the same one part-time technologist behind six separate Y1 platform builds totalling $446k (S1A $60k, S1B $150k, S2A $40k, S2B $80k, S3A $56k, S3B $60k);
• the same GM as partner-wrangler, supervisor of six program leads, and DASP tender owner;
• the same Alice Springs labour market — which DAS’s own past applications describe as having the highest population turnover in Australia and a sector workforce crisis.

An assessor who reads any two of the six spots this. Spray-and-pray is not the count; it’s the absence of a visible concurrency plan. That is fixable (see §6 coherence devices) and must be fixed in all six before submission.

The board member’s view

“You are asking me to authorise ~$20.7M of delivery obligations against a $197k equity base, with no resolution on what we do if three of them land.” The portfolio needs a board resolution covering: (a) authority to submit all six, (b) a pre-agreed acceptance hierarchy, (c) explicit authority to descope or decline at grant-agreement negotiation. Without (c) on paper, a multi-win outcome forces a panicked board meeting in the worst possible week. Get this resolution by ~17 June.

2. Win-combination matrix

Steady-state (Y2) figures. “New FTE” = DAS-employed only, excludes casual pools and partner roles.

Acceptance hierarchy (recommend the board resolve this order): 1. S1A → 2. S3A → 3. S2A → 4. S2B → 5. S1B → 6. S3B. Reasoning: S1A is the succession of the existing cohort and program (walking away strands the Leadership Group); S3A is 5-year recurrent revenue against the FY28 cliff and the strongest competitive position; S2A is cheap, clean, deliverable; S2B is the smallest, shortest, most NT-weighted consortium; S1B and S3B are the lottery tickets whose wins create the most danger.

Protective reality check: DAS will realistically win 1–3, concentrated in the individuals — which ARE deliverable together. The danger scenario is not six wins (the department won’t do it); it’s a consortium win landing on top of two individual wins with no ops layer. Plan for that specific case.

3. Per-concept prosecution

S1A — Speak Strong NT. VERDICT: WOUNDED → CREDIBLE after budget surgery

Exposures:

1. It wears the Speaking Up underspend. This is the literal successor application to a grant currently running wages $87k against a $139k budget with a $57k surplus. “You underspent $580k over three years; why fund you $2.5M over five?” is the first question. The concept’s mitigation (Mar 2026 restructure) is only as good as its evidence — if there is no delivery run-rate data from Mar–Jun 2026 (sessions held, PAG attendance, named coordinator in seat), the restructure claim is hollow. Assemble that evidence this week or the Criterion 2 counter has no spine.
2. Remote travel under-costed ~3x. $20–30k/yr to serve Tennant Creek (1,000km round trip) plus “10+ remote communities by Y5”. Bottom-up: one 3-day, 2-person Tennant trip ≈ $1.8–2.3k (vehicle + 2 nights × 2 pax at Tennant rates + incidentals). A monthly Tennant presence alone ≈ $22–28k/yr. Add 30–40 remote community visits/yr by Y4–5 at $1.5–4k each = $60–130k/yr. DAS’s own Safeguarding proposal — its most professional costing — budgeted $100–120k/yr remote travel for an NT-wide footprint. Assessors with NT experience (and state officials are in the DPSC loop) catch under-costed remote travel instantly; it reads as “has never actually done this”.
3. Peer wages vs the promise. $58k/yr peak ≈ ~1,050 casual hours/yr ≈ 20 hrs/week across the entire peer workforce — against outputs of “peer-led delivery at ≥60% of sessions by Y5”, 150–200 completers/yr, PAGs in two towns, and co-facilitation in 10+ communities. The maths doesn’t hold.
4. FN partner subcontract line vs the on-country promise. $60k/yr peak buys roughly 0.4 FTE of NPYWC/Tangentyere-grade partner time plus interpreting — not co-delivery and cultural governance across 10+ communities.
5. Regulated-activity (DSI Act) costs asserted as “inside program ops” but not visibly costed.

Fixes: re-cost travel bottom-up using the Inclusion NT quote format (trips × unit costs) — expect +$40–60k/yr from Y2, funded by descoping the community count from “10+” to 6–8 by Y5 and trimming the Y4–5 tech line; lift peer wages or cut the 60% peer-delivery KPI to 40–50%; lift partner subcontracts or shrink the on-country promise to match; add a recruitment/relocation line (~$5–8k per role — DAS has never budgeted one and cites recruitment as a kill risk in the same document); add an explicit CoC/NSDS cost line. Lead Criterion 2 with the restructure evidence and the structural fix (team of 2.5 FTE + casual pool replaces the sole-worker model that caused the underspend).

S1B — Self-Advocacy Workbench. VERDICT: WOUNDED — submit only the descoped shape

Exposures:

1. 8 named partners by 2 July is fantasy. Warm: DCLS, IADA, WOSCA, SACID (needs repair first), NPYWC (long approval cycle, also being asked to anchor S3B). Cold: VALID, Speak Out, CID — three established east-coast orgs being cold-called via a DANA introduction to join an Alice Springs-led national consortium, in the same month they are writing their own DPSC applications (CID and VALID will certainly bid their own Stream 1). Probability all 8 commit in-principle in 3 weeks: low. Naming partners without written in-principle agreement is the one move that crosses from optimistic into misleading — do not do it.
2. Product duplication visible to the same panel. S1A’s six modules (Know Your Rights / Speak Up at Your Plan Meeting / Ask for a Review / Make a Complaint / Choose and Direct Your Supports / Decide With Support) and S1B’s six pathways (Understand My Plan / Get Ready for My Plan Meeting / Ask for an Internal Review / Make a Complaint / My Supported Decision Plan / Speak Up at Work-School-Service) are the same product described twice. Same stream = same assessors, reading both side by side. Unmanaged, this poisons both applications (“double-dipping”).
3. “In-house production AI capability exists today” is unverifiable. DAS has no product in market an assessor can look at. The claim reads as aspiration attached to one part-time person (see §4).
4. Y1 not staged. $1.135M out of the gate — platform build + 5-region design panels + 2 coach cohorts simultaneously, from an org that took 3 years to underspend a 1-FTE project.
5. Senior salaries under-rated. 4.0 FTE for $505k ≈ $126k/FTE loaded. A national-consortium Program Director and a national product/tech lead at ~$100–110k base each will not recruit. This is how Speaking Up failed — budget a salary the market refuses, then underspend.

Fixes (the descoped shape): submit with 5–6 named warm partners (DCLS, IADA, WOSCA, SACID-if-repaired, NPYWC-if-landed; otherwise 4) + the §3.2 18-month national scale pathway via DANA explicitly invoked for QLD/WA/ACT/Vic/NSW — the Guidelines hand you this mechanism, use it instead of cold partners. Stage Y1 to $900k (platform + NT/SA pilot only; national launch Y2). Re-rate PD and tech lead to $165–185k loaded each, funded from the partner-flow line of the unfilled slots. Split the tech line into product-lead salary + named external development contract + escrow/handover deliverable (see §4). Add the cross-app declaration: “should both DAS Stream 1 applications be funded, digital development consolidates to a single build; the duplicated cost ($60k) is surrendered or redirected at agreement negotiation.” Address plan-document privacy explicitly (participants upload NDIS plans — sensitive health information held by a 7-person org; name the hosting, consent, breach-response and independent security audit arrangements, costed).

S2A — Open Doors. VERDICT: CREDIBLE

Exposures:

1. The ACCO partner is load-bearing and unconfirmed. The cultural-safety review domain — one of the five domains and the Closing the Gap PR3 story — depends on a partner not yet approached. If it isn’t locked by ~25 June the criterion text overclaims.
2. Assessor-pool churn. 10 work-capable casual assessors recruited from the lived-experience cohort, sustained over 3 years, in a casual model, with health and fatigue realities the concept itself names. The over-recruitment mitigation is right; the budget has no recruitment/onboarding line to pay for it (the Inclusion NT quote precedent budgeted $3k/role).
3. Demand side soft: ~50 organisations/yr opting in to be reviewed, free and recognition-incentivised — plausible in Alice, thinner in Tennant Creek. Y1 pilot target of 15 is honest; keep it.
4. Travel ($20–24k/yr + lease) is the closest-to-plausible of the six but still tight for a 50-review/yr two-town circuit; re-check with 8–10 Tennant trips/yr priced in.

Fixes: ACCO in-principle (Tangentyere or Anyinginyi) within 2 weeks; if it fails, restructure the cultural-safety domain as “co-designed with First Nations assessors and Leadership Group members, with ACCO partnership formalised in Y1” — weaker but honest. Add $8–10k/yr recruitment/onboarding. Otherwise submit as is — this is the portfolio’s cleanest application and the cheapest coverage buy for the department.

S2B — Inclusion Engine. VERDICT: WOUNDED

Exposures:

1. The “multi-state” claim hangs on SACID + a cold Partner E. SACID is the org DAS underdelivered a sub-grant to ($288 wages in 8 months against the CB&PLG MOU — and the “employ a person with ID” condition at risk). If the department phones SACID as a de facto referee, today’s answer hurts. Repair before naming; if repair fails and Partner E doesn’t land, the application is NT-only — at which point the title, geography selection and half of Criterion 1 are wrong. Keep an NT+remote-SA3s fallback framing drafted.
2. AI Easy Read is not whitespace. Photosymbols EasyMaker, Easy Read Toolbox, readeasy.ai already exist (landscape scan flags this in red). The differentiator is the paid lived-experience QA veto + remote/FN priority + in-house capability transfer — Criterion 1 must argue that explicitly or an informed assessor scores it as “funding a product that exists”.
3. The FN story is deferred. First Nations Plain English module “pending”, ACCO partner pushed to Y2 — against a program-level 10% First Nations delivery priority. Weakest FN angle of the six.
4. Throughput promise (250–300 orgs through 10-week sprints) is partner-delivered at $82–87k/yr/partner including panel wages and travel — roughly 0.5 FTE facilitation per partner for 20–30 orgs/yr. Arguable, but there is no acquisition strategy or marketing line for filling those cohorts.
5. Program Director at $140k total cost (~$115k base) — under-market again.

Fixes: Scott’s SACID repair call is the first partner action of the three weeks (it unlocks S1B too); pre-identify the WA/SA DANA fallback for Partner E now, not later. One paragraph in Criterion 1 naming the existing AI Easy Read products and the three differentiators. Pull one ACCO conversation forward into the application window even if only as “in negotiation”. Re-rate the PD. Add a cohort-recruitment KPI and a small activation budget so the 250-org number has a mechanism.

S3A — CDIS. VERDICT: CREDIBLE — best strategic bet, two lines must be fixed

Exposures:

1. The differentiation is under-costed. The whole case against Gateway duplication rests on first-language products and a physical outreach circuit. The budget funds neither at promise level:
   • Language line $38–46k/yr for production + cultural review across four languages. Bottom-up estimate: one topic pack professionally translated, recorded with paid community voice talent, and culturally reviewed ≈ $3–6k per language → $12–24k per 4-language pack. The SLA (“priority packs in ≥4 first languages”, 12 packs/yr) implies $70–140k/yr if even half the packs get full treatment. Under-costed 2–3x against its own service levels.
   • Travel $24–28k/yr for “minimum 6 community visits/quarter” + monthly Tennant presence ≈ $1k/visit all-in. Under-costed ~2x (a Tennant monthly presence alone consumes the line). An assessor who believes the gap but not the budget marks Criterion 3 down on the exact thing the application is for. This is the known remote-delivery red flag.
2. Identified-role recruitment (1.0 FTE First Nations IAR Officer in Alice Springs) — mitigations are written (trainee pathway, casual pool) but the budget should show the trainee money.
3. DAIS additionality must be airtight and proactively shared with NT Office of Disability — the double-funding question is the one eligibility attack with teeth.
4. FN governance sitting fees are folded into small lines; DAS’s own Safeguarding proposal benchmarked paid FN advisory structures at $120k/yr. A cultural review panel with sign-off rights costed at loose change reads as garnish.

Fixes: EITHER lift language + travel by ~$45–65k/yr combined (fund from the Y4–5 evaluation shape and the tech run-rate) OR soften the SLAs to “top 4 priority packs per year in 4 languages, expanding with cultural review capacity” and 4 visits/quarter. Do not submit the current promise on the current money. Cost the cultural review panel visibly ($30–50k/yr). Everything else — Gateway differentiation (A.11.1), Cat B logic, the §8.1 gap position — is genuinely strong; protect this application’s drafting time accordingly.

S3B — Remote Reach. VERDICT: NOT CREDIBLE AS SCOPED — salvage to minimum-viable

Exposures:

1. The partner table is a wish list. Confirmed today: DCLS + IADA (keen), Inclusion NT (warm). Unconfirmed: NPYWC, Congress, Anyinginyi — i.e. every ACCO, in a concept whose entire premise is ACCO-delivered access points and First Nations governance. A remote First Nations IAR consortium where no First Nations organisation has agreed is not a consortium; it’s a diagram.
2. The anchor partner is the incumbent competitor. NPYWC holds the closest existing grant to this concept (IAR 2025-26, “Supporting Information for Anangu with Disabilities”, $1.11M, WA/SA/NT, malpa model) and will be writing its own DPSC renewal during the exact window DAS asks it to anchor a DAS-led overlapping bid — while also being asked to join S1B. The realistic answer in 3 weeks is “in principle, maybe, after our board cycle”. The concepts’ own logic for the rejected employment alternate applies verbatim here: if NPYWC leads remote First Nations IAR, DAS should partner — not lead. The assessor (and DRC Vol 9) leans the same way: fund ACCOs to deliver.
3. Sub-grant machinery costed at $28k/yr to administer 5 sub-grant deeds worth ~$450k/yr — partner screening, quarterly per-partner reporting, staged payments, withholding, variance. Realistic admin is 5–8% of flow plus audit. DAS has been on the receiving end of these chains; it has never run one.
4. Tri-state compliance and travel: 3 jurisdictions’ WWVP/child-safe regimes, network practice forums, NPY-lands trips at $3–5k each — against $35–40k/yr DAS-side travel. Language pool $35–55k/yr for ≥6 languages plus Auslan commissioning (estimate $1.5–2.5k per finished short video before talent and review) — under-costed ~2x.
5. $1.17M/yr ≥ DAS’s entire current revenue, flowing through the org with the thinnest delivery evidence of the six.

Fixes (minimum-viable shape): descope geography honestly — NT-anchored network (Top End via DCLS/IADA, Centre via DAS-adjacent partners, Barkly aspiration), with the tri-state/cross-border layer expressed as “subject to NPYWC partnership, in negotiation” rather than claimed; name ONLY partners with written in-principle agreement; include an explicit complementarity paragraph re NPYWC’s incumbent grant (MOU sought, no on-lands overlap); cost sub-grant administration as a visible 0.2–0.3 FTE; stage Y1 to ~$700–800k. Accept the Criterion 2 bleed and the lottery-ticket status; spend the least drafting time here (see §7 triage). Hard line: if no ACCO in-principle lands by 25 June, the application ships with zero named ACCOs and says so plainly — overclaiming here is the portfolio’s single biggest integrity risk.

4. The tech key-person prosecution (cross-cutting)

The claim, six times over: “in-house production AI capability”, “DAS can demo one”, “built and run in-house”, “the tech layer makes a 7-person org deliver like a 20-person one”.

The facts a sceptic assembles: the capability is one person; he is part-time; he is also the grant writer; DAS has no public digital product an assessor can inspect; six applications budget six separate Y1 builds ($446k total) of what is structurally one platform spine (guided pathway / self-assessment → AI-assisted draft → human QA gate → action plan/dashboard → DEX-aligned data); no budget names an external development partner, escrow arrangement, or handover deliverable; and the deepest products (Workbench, CDIS console) involve vulnerable users and uploaded NDIS plans — sensitive health information — with “security” buried in hosting lines and no data-governance narrative.

If he is unavailable for a quarter, every tech line in every funded program stalls simultaneously. No assessor will write that sentence, but a good one will think it, and a board member must.

Honest mitigations (apply to all six, mandatory for S1B/S2B/S3B):

1. One architecture, declared. Each application budgets its delta (content, localisation, accessibility audits), and all six carry the same sentence: digital components share a common DAS architecture; concurrent awards consolidate build costs at agreement negotiation. This converts the double-funding optic into an efficiency story.
2. Split every tech line into (a) product lead salary at market rate, (b) a contracted external development partner (“to be procured Q1 FY28” is acceptable; a named Australian firm is better), (c) an escrow/documentation deliverable in Y1 — architecture docs, handover pack, second contracted developer by Y2. Open-licensing is already promised from Y4; bring the code/infrastructure documentation commitment forward to Y1.
3. Reframe the Criterion 3 claim from “rare in-house AI capability” (unverifiable, key-person-shaped) to “working prototypes + dedicated funded product team + external delivery partner under DAS product direction, all assets escrowed and openly licensed”. Attach prototype evidence inside the AWP if the template allows.
4. Shrink the AI surface where it touches vulnerable users: the “AI rehearsal partner” should be scoped as scripted-scenario practice with human peer debrief, not open-ended generative chat; Easy Read AI stays drafting-only behind the paid reviewer veto (S2B already has this right — replicate the framing).
5. Data governance paragraph + costed independent security audit in every application whose product touches personal information (S1A, S1B, S3A, S3B).

5. Personnel and the double-counted workforce (cross-cutting)

What’s clean: every budget funds new, project-dedicated roles. No application names Lukas or Scott as funded personnel; the §5.4 usual-responsibility trap is avoided on paper. Keep it that way — audit every draft so no named existing staff member appears as delivery capacity in more than one application.

Where the maths actually breaks:

1. The Leadership Group is counted six times. The same ~30 people are the §5.1 mechanism in all six AND the recruitment pool for S1A’s 25–30 peer mentors, S2A’s 10 assessors, S3A’s co-design pool, and S2B’s NT reviewers. Alice Springs does not contain a second cohort. Fix: a single DAS Lived Experience Workforce framing used identically in all six — one pool, multiple deployment streams, per-person cross-program hour caps, an annual new-member recruitment target. This converts the double-count into the portfolio’s best story (a town-scale paid disability workforce pipeline) and answers the capacity question before it’s asked.
2. Five separate First Nations governance structures are proposed (S1A cultural governance group, S2A ACCO co-design, S2B FN reviewers, S3A sign-off panel, S3B governance group) drawing on the same finite set of regional cultural authorities. Fix: one properly paid, partner-appointed DAS First Nations Governance Group serving all DAS DPSC activity, declared identically in each application, costed at Safeguarding-proposal benchmarks ($40–60k/yr portfolio-wide minimum), with program-specific sub-panels only where cultural protocols require.
3. Nobody manages the managers. Six program leads would report to one GM who also runs NDAP, the DASP transition and the board. No budget funds an operations/management layer; Speaking Up’s own budget precedent (“project worker + management support”) shows activity-attributed management fractions are fundable. Fix: 0.1–0.15 FTE management-support allocations in at least the three largest budgets, and a Criterion 2 supervision narrative (GM + board + per-program lead) in all six. If 3+ wins land, the first negotiation-stage move is converting those fractions into an ops manager.
4. Six senior recruits, one labour market. The portfolio’s own risk registers cite Alice Springs recruitment as a kill risk while collectively assuming 17 hires in 12 months with zero recruitment/relocation budget anywhere. Fix: recruitment/onboarding lines ($5–8k/role) in every budget; remote-hire (work-from-anywhere) explicitly allowed for the product roles; staggered start dates written into AWPs.

6. Delivery-history optics (cross-cutting)

What the funder can see: Speaking Up underspend in DEX/acquittals (wages $87k vs $139k budget, $57k surplus); the SACID sub-grant ($288 wages in 8 months) via SACID’s reporting chain; GO6984 submitted with an empty project plan table (same department family). NT 5-Year ($0 wages against a funded 0.5 FTE) is NT-funded — invisible to DoHDA directly, but state officials sit inside DPSC planning.

Exposure ranking: S1B worst (“underdelivered $193k/yr locally, wants to lead $1.2M/yr nationally”), then S1A (direct successor — but best armed to answer), S3B (biggest ask, thinnest evidence), S2B, then S3A/S2A (different activity classes; least exposed).

The honest counter (use everywhere, once, briefly):

• The distinction that holds: DAS delivers its recurrent services reliably (NDAP since 2017, NDIS Appeals since 2016, NT DAIS, School Advocacy — continuous Commonwealth reporting) — its weakness was sole-worker project structures in a brutal recruitment market. Every new budget replaces sole-worker structures with teams + coordinators + casual pools and realistic Y1 ramps. Name that lesson explicitly; assessors reward organisations that can describe their own failure mechanism.
• One sentence of pre-emption in Criterion 2, not a confession paragraph: acknowledge the ramp-up shortfall, point at the March 2026 restructure, cite the post-restructure run-rate. This requires the run-rate to exist — assembling Mar–Jun 2026 delivery numbers is a week-1 task. If they don’t exist, say less and rely on the structural-fix narrative.
• Never claim throughput track record. Claim continuity, lived-experience employment (the co-facilitators-into-paid-work line is the best single sentence in the corpus), co-design depth, and recurrent-service reliability.
• SACID is both a wound and a witness. Do not name SACID as a partner in S1B/S2B until the repair conversation has happened — the department may treat partners as informal referees.

7. Budget realism — pattern verdicts

8. Partner herding — the 3-week reality

The full ask list: ~11 distinct organisations, ~15 separate roles, several orgs asked 2–3 times (DCLS and IADA appear in all three consortium applications; SACID in two; NPYWC in two plus an S1A subcontract mention; Inclusion NT in two) — all during the exact weeks those organisations are writing their own DPSC applications.

The only strategy that works: one conversation per organisation covering every role.

Descope triggers (pre-decide now): SACID unrepaired by 20 Jun → S2B goes NT+Partner-E-or-fallback and S1B drops to the warm-4. No ACCO in-principle for S3B by 25 Jun → S3B ships with zero named ACCOs and honest “in negotiation” language. NPYWC not landed by 25 Jun → S3B drops tri-state claims; S1B proceeds without (WOSCA + identified role carry the FN angle, weaker but real).

9. Recommended portfolio posture

All six submitted, in three tiers, with portfolio coherence engineered in.

• Tier 1 — full-dress (protect drafting quality): S1A, S3A, S2A. Highest win probability, deliverable in combination (the AMBER-GREEN target outcome), and they carry DAS’s strategic needs (succession, recurrent revenue, coverage).
• Tier 2 — descoped consortium shapes: S2B (4–5 named partners, SACID contingency, AI-Easy-Read differentiation paragraph), S1B (5–6 warm partners + §3.2 scale path, staged Y1 ~$900k, market-rate seniors, external dev + escrow).
• Tier 3 — minimum-viable: S3B (NT-anchored descope, honest partner status, costed sub-grant admin, leanest drafting from templates). It stays in the portfolio as a ticket and as signal of network ambition; it gets no scarce hours that Tier 1 needs.

Coherence devices, mandatory in all six:

1. Identical disclosure-and-portfolio paragraph: number of applications; the logic (individual apps = place-based remote coverage; consortium apps = sector-shared infrastructure); independently deliverable; no funded personnel shared across applications; concurrency managed per the Risk Management Plan.
2. A concurrent-award risk in every Risk Management Plan (staggered starts, consolidated digital architecture, negotiation-stage descopes). Self-awareness here is the single cheapest credibility purchase available.
3. One budget workbook, one rate card, one set of org facts (origin year standardised, FY24-25 financials, ABN/CGH ID) so no two applications contradict each other — the corpus’s history of self-contradiction is a known defect.
4. Single Lived Experience Workforce and single FN Governance Group framings (§5).
5. Board resolution: submit six + acceptance hierarchy (S1A → S3A → S2A → S2B → S1B → S3B) + authority to descope/decline at negotiation.

The one truly-indefensible flag (within the six-lock): submitting S3B — or any consortium application — naming partners who have not given written in-principle agreement. That is the line between an ambitious portfolio and a misleading one, and it is also the failure mode (hollow consortium tables) that sank GO6984. Everything else in the portfolio is fixable with descopes and re-costing; this one is integrity.

10. The 3-week writing plan

Capacity honestly stated: the drafting is the cheap part — the concepts are 80% of the intellectual content and AI compresses criteria text fast. The real binds are (a) partner conversations (human, unparallelisable), (b) verification chores, (c) Scott’s review bandwidth, (d) form hygiene across six smart-forms — the exact failure mode of GO6984 — and (e) the DASP tender breathing down the schedule. The plan protects (e) by finishing DPSC on 30 June, not 2 July.

Week 1 — Thu 12 → Wed 18 Jun: decisions, partners, templates

• Scott: SACID repair call (day 1–2); DCLS+IADA combined meeting; NPYWC combined approach with MOU-in-principle template; Inclusion NT, WOSCA, Anyinginyi, Congress, Tangentyere approaches; DANA intro requests (VALID/CID/Speak Out/Partner E). Board paper + resolution (submit six, hierarchy, descope/decline authority) by 17 Jun.
• Lukas: build the template kit once — org-capability block (Criterion 2 core), lived-experience workforce block, FN governance block, delivery-history pre-emption sentence + evidence pack (Mar–Jun 2026 Speaking Up run-rate — assemble or confirm absent), Gateway-differentiation blocks (S3 pair), disclosure/concurrency paragraph, risk-plan library (10 standard + per-app specials incl. concurrent-award risk), AWP skeleton, single budget workbook (FY28-escalated SCHADS rate card, on-costs, remote loadings, recruitment lines, bottom-up travel calculator, 6 tabs). Verification chores: SA3 codes (ASGS Ed 3), FY24-25 financials, bank verification <6 months, origin-story standardisation, partner ineligibility screens as names confirm (Redress/WGEA/NDIS Commission/s73ZN — ~11 orgs).
• Both: lodge any Grant.ATM eligibility de-risk questions by Fri 20 Jun (Q&A closes ≈25 Jun).

Week 2 — Thu 19 → Wed 25 Jun: Tier 1 complete, partners locked

• Lukas: full drafts of S1A, S3A, S2A — criteria + AWP + budget + risk plan — with the §3/§7 budget surgery applied (travel re-cost, language re-cost, peer wages, recruitment lines). Start consortium shared sections (governance, sub-grant mechanics, partner tables pending).
• Scott: review pass 1 on Tier 1 (criteria against the Guidelines vocabulary, honesty check on every output number); chase outstanding partner in-principles. Partner tables lock Wed 25 Jun — whoever is in writing is named; everyone else becomes pipeline/scale-plan language. Descope triggers from §8 fire automatically.

Week 3 — Thu 26 → Tue 30 Jun: Tier 2/3, QA, submit

• Lukas: S2B and S1B drafts (descoped shapes) from templates + locked partner tables; S3B minimum-viable draft last (target: one day, template-heavy).
• Second reader (board member or paid external proofreader — $2–3k of insurance): the GO6984 checklist on all six: every table populated, no instruction text left in responses, character counts, unique titles, attachments <2MB, typo pass, no cross-app contradictions (run a diff on org facts), budget tabs reconcile to forms.
• Both: form entry and submission of all six Tue 30 Jun — 48-hour buffer, submit once only (latest-submitted-wins rule means a re-submission replaces, so there are no do-overs). No day-of-deadline heroics; that habit produced the historical defect list.
• 1–16 Jul: DASP tender, reusing the template kit (org capability, delivery history, FN governance blocks all transfer). This fortnight is why Tier 3 gets no extra hours: one day of S3B polish is worth less than one day of DASP prep, and DASP is existential.

If capacity binds anyway (the triage order)

Thin in this order: 1. S3B (already minimum-viable — it can survive on concept-doc compression), 2. S1B (its content reuses S1A’s heavily; cut to the descoped core), 3. S2B. Never thin: S1A, S3A, S2A, or the QA pass. If a whole application must effectively be sacrificed to protect the DASP runway, it is S3B — submitted lean and honest rather than withdrawn, since six is locked and a lean-honest application costs little and keeps the network signal.

11. The single biggest deliverability hole

Six budgets quietly assume exclusive use of the same four finite inputs: one part-time technologist, one ~30-person lived-experience pool, one GM’s bandwidth, and Alice Springs’ worst-in-class labour market — and no budget anywhere funds the operations/management layer that even a 2–3-win outcome requires. Every additional win currently degrades every other program. The fixes are cheap and mostly textual (one-architecture declaration, single workforce and governance framings, management-support fractions, recruitment lines, the concurrency risk in every risk plan) — but they must be applied to all six before submission, because the disclosure question guarantees the assessor reads the portfolio, not the application.