Skip to content

ADR 0008: COTS-First MVP Strategy

Date: 2026-05-08 Status: ACCEPTED Supersedes (in part): ADR 0002 — Hardware Tiering (BOM targets only — the 97/3 architecture stands), ADR 0006 — Tiered Service Pricing (specific tier prices treated as aspirational pending real cost data). Invalidates assumptions in: docs/00_business/strategy/economics.md, docs/70_internal/financials/strategic-economics-blueprint.md, docs/70_internal/team/co_founder_brief.md — funding tables and capital story.

Context

The economic and investment-series planning done in Q1 2026 was based on assumptions that did not survive a 2026-05-08 reality check:

  • Hardware BOM (Member < $15, Leader < $80 at 250k+ unit volume) was modelled before getting real vendor quotes at MVP volume.
  • Adoption curves (path to 2.5M head, 0.4%/mo churn, 77% pre-purchase mix) were modelled without a single field pilot.
  • Funding ladder ($450K Seed → $2M Series A → $6M Series B, with three internally-inconsistent variants across the docs) implied a level of milestone certainty the project did not have.
  • Team scale assumed in the model was not committed.

Those assumptions, together, produced economics that read as confident but rested on weak ground. Continuing to plan around them — and to fund-raise on them — would compound the risk.

Decision

The MVP and the next 6–12 months use commercial off-the-shelf (COTS) products as much as possible in place of custom-designed hardware. Concretely:

  • Member node: built on a COTS LoRa devkit (RAK3172-class or equivalent) rather than a custom PCBA. Battery life and power-budget assumptions are validated against the actual COTS module rather than against a target spec.
  • Leader node (Cowbell): built around a COTS satellite gateway (Blues Notecard-class or equivalent) rather than a custom NTN integration. Connectivity cost is whatever the COTS module's bundled plan charges, not a model.
  • Cloud / platform: built on managed services where possible (no premature lock-in to self-hosted infra unless data-sovereignty requirements force it — see R-09 for the Blues / Notehub trade-off).
  • Custom hardware optimization (driving Member toward < $20, Leader toward < $80 at scale) is deferred until post-validation, when both BOM targets and pricing can be tuned against real numbers.

The strategic intent of the COTS path:

  1. Validate market demand with minimal capital outlay. A real customer paying for a COTS-based deployment is worth more than a model.
  2. Develop platform IP (firmware, edge AI, cloud, data product) without committing to custom hardware too early. The platform is the moat, not the silicon choice.
  3. Raise investment from a position of strength post-validation, with real pilot economics, real customer data, and a real BOM gradient that justifies committing capital to custom hardware.

Consequences

What's now on hold (not deleted, on hold)

  • The "< USD 50 / 3-year all-in" headline target in the README and economics doc — re-frames as a post-validation goal, not a current commitment.
  • The 3-tier service pricing in ADR 0006 ($6 / $12 / $18 per year) — directionally correct, but specific numbers are aspirational pending real cost data from COTS deployments.
  • The funding ladder in the strategic-economics-blueprint and the co-founder brief. Capital strategy will be re-derived after the validation phase produces real unit economics.
  • The Member < $15 / Leader < $80 BOM targets in ADR 0002. The 97/3 architecture itself is unaffected; only the cost-targets are deferred.

What changes immediately

  • Unit economics in the MVP will be materially worse than the model — likely 5–10× the original Member-BOM target on the COTS path. This is the deliberate cost of validating market before committing custom-hardware capital.
  • Partner shortlist priority shifts. COTS-friendly vendors (Blues, RAK Wireless, GlobalSat) move up. Custom-silicon and ODM paths (T2M Semi, Racyics, direct Quectel ODM) defer until post-validation.
  • The next-milestone selection (the gating Phase 0 deliverable in proj/roadmap.md) is now scoped to a COTS-based market-validation engagement — a paying pilot or a tightly-scoped commercial trial — not a custom-hardware demo.
  • The Blues partner conversation (meeting week of 2026-05-11) becomes more important — it's now a candidate for the Leader-node backhaul in the COTS configuration.

What survives the reset

  • The architecture decisions — 97/3 hierarchy (ADR 0002), single-mode Star with mesh-capable hardware (ADR 0007), hybrid connectivity strategy (ADR 0005), platform-as-OS business framing (ADR 0004).
  • The brand and voice work (brand strategy, voice ES).
  • The regulatory and partner research — ANII positioning, EUDR readiness, the partnership-matrix intelligence.

Open Items

  • Define what "validated" means before this ADR can be retired:
  • Which metrics? (signed pilots? paying customers? measured retention? measured Star-topology stray rate? measured COTS unit cost?)
  • At what scale? (head count? number of distinct customers?)
  • Over what time horizon? (3 months of operation? 6? a full biological cycle?)
  • Pick the COTS Member candidate. RAK3172 is the placeholder; verify it survives the Star-topology power budget from ADR 0007.
  • Pick the COTS Leader candidate. Blues Notecard for Skylo is the leading candidate but introduces the Notehub data-sovereignty trade-off (R-09).
  • Update the partnership matrix with a COTS-friendly column or annotation so the shortlist is visually obvious to future readers.
  • Re-frame external pitch material. Decks, briefs, and investor-facing surfaces should lead with "validating market with COTS, custom hardware as the post-validation play" rather than the original "$8M ladder to 2.5M head" framing.

Notes

  • This decision is firm but not eternal. If validation produces evidence that the COTS path cannot reach the cost or power targets needed for mass adoption, a follow-up ADR will revisit and may pivot back toward custom hardware sooner than otherwise planned.
  • Documents that captured the original economic modelling are preserved with banners pointing at this ADR rather than rewritten — the institutional memory of what was modelled and why has independent value.