THE INGO LAYER IS GONE.
WE'RE WHAT REPLACES IT.

The bottom-up math

luma is an AI-native services company with two primary revenue lines — government contracts (multi-year MoH master service agreements, recurring) and pharma F100 RWE buyers (cohort data API, high-margin) — with two smaller secondary streams (public-health surveillance APIs, academic research licenses). Pharma revenue is what lets the government-services tier scale without donor backing. The math is bottom-up, country-by-country.

What $1.5M buys

5-person team, 18 months. The plan is paced to the partnership signal we already have, not invented backwards from a target raise.

From wedge to operating system, layer by layer

The chatbot is the wedge — what gets us into the ministry. The ten-year company is the rest of the INGO function, replaced one layer at a time. Each layer adds to the contract value, deepens the pharma data asset, and pushes the cumulative "% of INGO function luma replaces" number from ~30% today toward 85% at full stack.

The pricing trajectory. $250 → $400 → $500 → $600 → $700 per CHW per year as layers ship. At a 50,000-CHW network across four countries by year three, that's the difference between $12.5M ARR (workflow only) and $35M ARR (full stack) on the ministry side alone. Pharma RWE is additive on top, on the same data, growing in step. Every layer makes both sides more valuable.

Order of operations is deliberate. Don't build all six at once. Wedge → depth → upsell → moat. Each layer is a separate buyer conversation but the same install base, the same data pipeline, and the same engineering team — which is why this is a defensible expansion story rather than a six-product diversification.

5 hires, 18 months

Founder builds (full-stack + clinical-product fluency). First five hires unblock the GTM and deepen the platform.

Four buyer types, four price points

Same platform, four pricing surfaces. The unit-economics of the pharma layer (high contract value, low marginal cost per query) are what make the workflow layer affordable for ministries.

What we're up against

No single competitor sits in our position — workflow tool that is also a structured-data layer for pharma RWE. Closest analogs are split between workflow-only (CommCare, Ona, Medic Mobile) and data-only (Datavant, Sermo, Truveta).

Defensibility: the moat is the operational data accumulated from CHW interactions over time. Once a ministry adopts luma, switching costs are high (protocol corpus, language model fine-tuning, district workflows). Pharma buyers are sticky because the data refreshes weekly — replacing it requires a five-year DHS cycle.

What can go wrong

Honest list. The mitigation row is what we're already doing.

Four structural shifts converging

1. The off-ramp moment. The 2025 US foreign-aid restructuring forced a forty-year-overdue conversation about recipient-country independence from donor cycles. Ministries inherited the responsibility without the INGO implementation layer that used to deliver it. Software is the only substitution path that fits the budget.

2. WhatsApp is the universal SSA messaging layer. 80%+ smartphone penetration in Lesotho's CHW workforce. WhatsApp Business API stable since 2022. Not a hypothesis — a measured fact in our pilot.

3. Sesotho/Bantu-family ASR has crossed the usability threshold. Whisper v3 + minor fine-tuning gives transcription quality good enough for clinical Q&A. Five years ago this was not possible at this cost.

4. Pharma RWE budgets shifted to LMIC populations. ICH E17 (2018) and FDA RWE framework (2023) accept LMIC trial sites. Pharma are now actively budgeting for African RWE access — historically they could not. The same operational data the ministry layer generates is the asset pharma is now budgeted to buy.

Each on its own is a known trend; the intersection is a 5–7 year window. luma exists to occupy it before someone else does — and to make the resulting infrastructure government-contracted, not donor-funded.

The unit economics that turn aid-dependence into ministry-independence

Why this team can do this

Two co-founders. Sebastian Buck — technical product, public-health-fluent, prior operator experience in regulated software, eight weeks of CHW fieldwork in Lesotho with pre-existing ministry relationships. Vrinda Sood — [domain expertise placeholder — replace with Vrinda's real bio: education, prior shipped work, the specific edge she brings to luma]. The combination is unusual — most US health-tech founders don't speak Sesotho; most SSA public-health professionals don't ship production software. luma is being built precisely at that overlap.

The plan above front-loads two hires (country lead, pharma GTM) that complement the founders' coverage. Series A buys the third country lead and engineering depth.

$1.5M seed

18 months of runway. Closes the Lesotho ministry contract, executes the first pharma anchor, expands to Eswatini, hires 5 people. Series A target month 18 at $6–9M ARR.

Anyone who's read this far: sebastian@meetluma.org.