# Lighthouse Care: proposed provider decision pilot

This is a **proposal for a four to six week evaluation**, not a completed pilot or evidence of realised savings, faster decisions, safer care, forecast accuracy or customer demand. The initial buyer hypothesis is a CEO or COO of an NDIS provider operating several supported homes. A finance lead and a care or quality lead should review each decision together.

## Job to test

Before an executive or board review, decide what to change at a home, what each option means for cash and people, what prevents action, who owns the smallest safe next step, and which observable evidence would reverse the recommendation. Lighthouse Care should connect dated history, current operating constraints, explicit future assumptions and relevant external signals while marking each input as observed, synthetic, connected, assumed or unknown.

The public demonstration uses synthetic operating records and selected dated public sources. It has no connected provider systems and does not approve clinical decisions, assign shifts, reserve beds or obtain participant consent. Any phase using provider records requires agreed tenancy, role access, privacy, consent, retention and security controls before import.

## Three decisions

| Decision | Question to rehearse | Evidence and hard constraints to review |
| --- | --- | --- |
| Home viability and redesign | Retain, change occupancy or support delivery, negotiate property cost, stage a transition or close? | Reconciled home P&L and cash, paid versus billed claims, roster and wage allocation, resident choice, tenancy, workforce consultation, continuity, option dependencies and reversal trigger. |
| Plan cut and workforce gap | What service can still be delivered after a funding change or missing shift, and where does the gap land? | Plan version and effective date, funded versus delivered hours, claim and payment timing, named dated coverage, credentials, clinical review, care consequences and a reversible response. Do not treat annual price drift as a one-off plan cut. |
| Referral or provider exit transfer | Can a financially attractive intake actually be supported, staged or declined? | Participant choice and consent on the proposed start date, housing and household fit, clinician capacity, named staff and roster evidence, funding, inherited obligations and the cost of safe transition. |

## Proposed sequence

1. **Week 1 — baseline and case selection.** The provider identifies three representative decisions and records current preparation time, current review materials, unresolved questions and who can approve action. Start with synthetic cases if provider data access is not ready.
2. **Weeks 2–3 — evidence and reconciliation.** Finance checks that delivered and paid time, claims, collections, property costs and allocations reconcile within a documented scope. Care and operations identify missing rights, clinical, housing and workforce evidence. Record source dates and uncertainty.
3. **Weeks 3–4 — joint decision reviews.** For each case, compare doing nothing with feasible alternatives. Review financial and human effects together. Record hard stops, the smallest safe action with an owner and review date, and a measurable reversal condition. Save and reopen the same decision record.
4. **Weeks 5–6 — independent replay and buying decision.** A second reviewer recovers sources, assumptions and history, reproduces the material calculation and challenges a no-go case. The sponsor compares preparation time and decision quality with the baseline and decides whether to fund another planning cycle.

## Proposed success criteria

These are **targets to agree with a design partner**, not achieved results.

| Test | Evidence of value to collect |
| --- | --- |
| Finance credibility | Finance lead accepts the reconciled baseline and every documented difference in scope or timing before options are evaluated. |
| Care credibility | Care and operations reviewers confirm all material blockers in the three cases are visible; no hard gate is represented as permission to act. |
| Historical usefulness | Reviewers identify which dated historical observation changed a question, assumption or option. Unexplained associations remain labelled unknown. |
| Decision speed | Measure the existing preparation process first, then target a material reduction, provisionally 50%, in time to a review-ready pack. |
| Decision usefulness | Each case contains alternatives, money and human consequences, unresolved dependencies, an owner and review date for the next step, and an observable reversal trigger. |
| Reproducibility | An independent reviewer reopens the saved record and reproduces material calculations from its version, inputs and evidence. A matching digest proves reproducibility, not economic completeness. |
| Commercial pull | The sponsor requests a further planning cycle and agrees a paid continuation against a named budget. |

The pilot can support a claim of **faster, better evidenced provider decisions** only if those measures are observed. Claims about reduced harm, improved clinical outcomes, realised savings or predictive accuracy need separate longitudinal evidence. Competitive advantage remains a hypothesis until buyers compare this workflow with their existing systems and alternatives.