Biome.Begin map
Product Spec · v0.1
May 2026
AF R&D

Biome — Product Specification

Version: 0.1 (concept lock candidate)

Date: May 2026

Owner: Kristen Hall, Director of R&D, America's Future

Status: Pre-build. Concept doc paired with Editorial № 04.

Working name: Biome.
Alternates if Biome doesn't clear: Habitat · Inner Climate · Verdant · Tend.
Final pick locks before Phase 1.

1. One-line summary

A questionnaire-driven web app that treats the human body as a nested set of environments, maps personal inputs to each sub-environment, and surfaces honest, cited information about how those environments grow health or grow disease. Never prescriptive. Never commercial.

2. Positioning

| | |

|---|---|

| Category | Health literacy / personal environment platform |

| Posture | Information-first. Mirror, not coach. |

| Closest analogs | None directly. Adjacent: Examine.com (supplement reviews), Cronometer (nutrition), NHS Health Check, biological-age calculators. |

| Differentiator | The nested-environment framing + total absence of commercial entanglement + algorithmic mirror that shows its math. |

| Portfolio position | Standalone platform. Natural bridge to Whiteboard's life-skills and continued-learning tracks. |

3. Audience

1. Primary — Adults who want to understand what's affecting their health without being sold to. Especially: people exhausted by influencer wellness, people whose clinicians don't have time for the whole picture, people piecing together their own answers.

2. Secondary — Parents thinking about family-wide health environment. Natural bridge to Family Track.

3. Tertiary — Clinicians, coaches, and counselors who want a neutral, citation-first reference tool to share with clients.

Explicitly not the audience: people looking to be told what to buy or what protocol to follow. Biome will frustrate them by design.

4. Bright lines (the negative spec)

These are conditions of existence, not preferences. Encoded in tech stack, content policy, and legal structure.

1. No supplement sales, no affiliate links, no brand recommendations, no sponsorship from any wellness/supplement/device/clinic company. Funding model isolates the platform from commercial conflict.

2. No prescription. Biome explains; it does not instruct. The verbs are *describe, show, map, compare, link to source*. The verbs are not *take, do, start, try, optimize*.

3. No proprietary opaque score. If a number is shown, the math is shown.

4. No deterministic prediction. Trajectories carry visible uncertainty.

5. No hiding sources. Every claim links to primary research.

6. No third-party analytics on user health data. No advertising pixels. No data sale. No data sharing without explicit per-instance consent.

7. No replacing clinical care. Disclaimer is persistent and load-bearing.

8. No surveillance of children. If family mode ships, children's data only joins to a verified parent account; never independent profiles.

5. Information architecture

Landing
  ├── About / Why Biome / Bright Lines
  ├── Source Policy (public, indexed)
  └── Sign in

App (authenticated)
  ├── Onboarding
  │   ├── Disclaimer + acknowledge
  │   ├── Privacy posture explainer
  │   └── Questionnaire (sectioned, skippable)
  │
  ├── Your Map  ← landing screen
  │   ├── Eight systems visualized
  │   ├── Tap a system → its inputs, indicators, evidence
  │   └── "What's actually known" deep-links
  │
  ├── Your Trajectory
  │   ├── Modeled trajectory (1y / 5y / lifetime where modeled)
  │   ├── Sensitivity sliders ("what if I changed X")
  │   ├── Life-expectancy modeling (optional, framed)
  │   └── Methodology page
  │
  ├── Library
  │   ├── Diet families (8)
  │   ├── Modalities (8+)
  │   ├── Sub-environment deep dives (8)
  │   ├── Hidden inputs (microplastics, EMF, light pollution, sound, social)
  │   ├── Life stages (puberty → aging)
  │   └── Video library (interview-style, researchers not influencers)
  │
  ├── My Notes (encrypted)
  └── Account / Export / Delete

Persistent disclaimer band on all modeled-output screens. Sources drawer accessible from every claim.

6. Onboarding questionnaire

Design: modular, sectioned, skippable. Three minutes for essentials; twelve to fifteen for full picture. Quarterly re-take recommended; trend view exposed.

Section breakdown

| Section | Approx. time | Notes |

|---|---|---|

| Demographics & history | 2 min | Age, sex assigned at birth, height/weight (optional), family history, prior conditions, current meds |

| Sleep | 1 min | Duration, regularity, perceived quality, light environment, partner/snoring/pets |

| Stress | 2 min | PSS-10 (validated instrument) or short version |

| Nutrition | 3 min | Dietary pattern self-identify + FFQ-lite + water, alcohol, caffeine |

| Movement | 1 min | Frequency, intensity, modalities |

| Environment | 1 min | Geographic (auto-pulls AQI from EPA AirNow), occupational, household chemical context, smoking/secondhand |

| Hormones | 1–2 min | Cycle (if applicable), menopause status (if applicable), thyroid context, optional lab upload (Phase 3) |

| Modalities used | 1 min | Supplements, peptides, light, vibration, sauna, cold, hormone therapy, breathwork |

| Life context | 1 min | Loss, caregiving, financial pressure, social connection, sense of purpose |

All questions skippable. Each section can run independently. No section blocks another.

Validated instruments (use carefully)

  • PSS-10 for stress (public domain, well-validated)
  • PSQI for sleep quality (consider; longer)
  • Avoid PHQ-9 / GAD-7 in v1 — clinical screening adjacent to mental-health diagnosis is a category Biome shouldn't enter without clinical partnership.

7. The systems map (data model)

Eight nodes. Each node:

SystemNode {
  id: string                    // gut, sleep, endocrine, immune, ...
  name: string
  currentState: {
    estimate: 'thriving' | 'stable' | 'strained' | 'compromised' | 'unknown'
    confidence: 0..1
    drivers: InputId[]          // which inputs are pulling this rating
  }
  inputs: InputId[]             // inputs that affect this node
  trajectory: {
    horizonYears: 1 | 5 | lifetime
    direction: 'improving' | 'stable' | 'declining'
    uncertaintyBand: [low, high]
    modelSource: CitationId
  }
  indicators: Indicator[]       // what to look for; biomarkers if known
  module: ModuleId              // deep-dive content
  citations: CitationId[]
}

Eight nodes:

1. Gut & Microbiome

2. Sleep Architecture

3. Endocrine

4. Immune

5. Cardiovascular & Metabolic

6. Cognitive & Nervous

7. Structural (bone, muscle, fascia)

8. Exposure Layer (chemical, light, EMF, sound, social, psychological)

Cross-system links explicit: stress → cortisol → gut → immune → sleep → cognitive. The graph is the point.

8. The algorithmic mirror

What it is: A reflective model that takes user inputs and runs them against published epidemiological risk modifiers to produce a *conditional* picture of trajectory.

What it is not: A predictor. A diagnostic. A score.

Approach

  • Use peer-reviewed all-cause mortality risk modifiers from major cohort studies: Framingham, UK Biobank, NHANES, Nurses' Health Study, ARIC.
  • Use condition-specific risk factors from meta-analyses where strong (cardiovascular: ASCVD pooled cohort; T2D: ADA risk; etc.).
  • Model output: current environment snapshot + modeled trajectory + sensitivity analysis.
  • Sensitivity sliders are the killer feature: adjust an input → see modeled effect on trajectory.

Life expectancy estimate

Optional. Opt-in. Always framed as modeling exercise.

  • Use published actuarial modifiers (smoking ~10 years, severe insomnia ~3 years, etc.) from peer-reviewed sources.
  • Show as range with uncertainty bounds, not point estimate.
  • Permanent framing: "Based on the inputs you've shared, mortality models from epidemiology suggest a range of…"
  • Users can hide this feature in settings.
  • Companion view: "what changes most" — sensitivity ordering.

Methodology

Published page. Every coefficient, every source. If we cannot show the math, we don't show the number.

9. Content modules

Sub-environment deep dives (8)

One module per system node. Pattern: what it is → what it needs → what wrecks it → what the evidence actually shows → open questions → trusted reading list.

Diet families (8)

Mediterranean · Paleo · Pescatarian · Plant-forward / WFPB · Low-carb / Ketogenic · Autoimmune Protocol · Intermittent Fasting Patterns · Carnivore. Each: evidence base, mechanisms, studied populations, likely strengths, open questions. No ranking.

Modality library (8+)

Peptides · Red light therapy · PEMF & whole-body vibration · Stem cell therapy · Sauna & heat · Cold exposure · Breathwork · Hormone optimization. Each: what it is, mechanism, strongest evidence, what's hype, what's contested, what's unregulated.

Hidden inputs

Microplastics · EMF · light pollution · noise · social environment · loneliness · purpose loss · caregiver load. The inputs people forget to count.

Life stages

Puberty · pregnancy & postpartum · perimenopause · menopause · andropause · aging. Each stage = new environment.

Video library

Interview-style with researchers, not influencers. Vetted production partners only. No sponsorship.

10. Interactive features

| Feature | Phase | Notes |

|---|---|---|

| Sectioned questionnaire | 1 | Save partial progress; resume; quarterly re-take |

| Systems map | 1 | Static-input version |

| Sensitivity sliders | 2 | "What if I changed this" |

| Trajectory view | 2 | 1y / 5y / lifetime modeled |

| Life expectancy (opt-in) | 2 | With full framing |

| Compare environments | 3 | A/B two dietary patterns, two sleep profiles, etc. |

| Quarterly trend view | 3 | Shows changes over time |

| Private notes | 1 | Encrypted at rest, per-user key |

| Clinician-export summary | 3 | PDF; user controls what's included |

| Family mode | 4 | Parent-anchored, never independent child profile |

| Self-hosting option | 5 | For high-trust / high-privacy users |

11. Source policy

Every claim has a citation. Hierarchy:

1. Cochrane / systematic review

2. Meta-analysis

3. Large prospective cohort (Framingham, UK Biobank, NHANES, Nurses')

4. Randomized controlled trial

5. Smaller observational

6. Mechanism / animal model

7. Expert opinion (rare; tagged)

Where evidence is contested, both positions shown with relative strength.

"Hype watch" tag: popular claim, weak evidence. Used for many supplement claims, several peptide claims, and roughly half the cold-exposure folklore.

Annual content audit. Citations re-verified. Modules updated. Audit log public.

12. Disclaimers (the load-bearing pattern)

  • Onboarding: full disclaimer modal + explicit acknowledge before any input is collected.
  • App header band: persistent on all modeled-output screens: *"Information, not medical advice. Not a substitute for clinical care."*
  • Decision-point reinforcements: repeated at the life-expectancy view, modality pages, dietary library.
  • Source policy: the disclaimer's enforcement mechanism. If we cite, we mean it. If we can't cite, we don't claim it.

13. Tech stack proposal

Aligned to existing portfolio for build velocity and reusability.

| Layer | Choice | Rationale |

|---|---|---|

| Framework | Next.js 16 | Matches Project Exodus, Whiteboard direction |

| Language | TypeScript | Portfolio standard |

| Styling | Tailwind v4 | Portfolio standard |

| DB | Postgres (Neon via Vercel) | Matches Exodus |

| ORM | Prisma | Matches Exodus |

| Auth | TBD — research Lucia or Clerk before defaulting to NextAuth | Exodus had NextAuth pain at session end |

| User-content encryption | E2EE pattern from Halo | Notes encrypted at rest, per-user key |

| Charts | Recharts | Trajectory visualizations |

| Air quality data | EPA AirNow API | Public, free, authoritative |

| Water quality data | EWG Tap Water Database (API or scrape with permission) | Best available consumer-facing source |

| Research lookups | PubMed via NCBI E-utilities | Cached to internal DB to avoid rate limits |

| Hosting | Vercel | Portfolio standard |

| Repo | Private, researchpastfuture/biome | Portfolio policy |

| Deployment license | Publicly licensed | Portfolio policy |

| Domain | Reserve biome.americasfuture.net (or chosen name) before public concept share |

14. Data & privacy posture

Health data is the most sensitive category. Treated accordingly.

  • All user data encrypted at rest with per-user keys.
  • No third-party analytics. No advertising pixels. No data brokers.
  • Self-hosting option in Phase 5 for users who want full control.
  • Family mode: children's data only ever joined to a verified parent account. No independent child profiles.
  • Export & delete buttons prominent in account screen.
  • Privacy policy in plain language, no dark patterns.
  • Data residency disclosed.
  • No data sale, no data sharing without explicit per-instance consent.
  • Inherits security instincts from broader portfolio: auth tokens treated as crown jewels, secrets in vault (consider Conduit pattern when Conduit is live).

15. Build phases

| Phase | Scope | Definition of done |

|---|---|---|

| 0 — Concept lock | Name finalized · brand identity · source policy · legal framing | Name registered · domain reserved · source policy v1 published |

| 1 — Questionnaire + 3 modules + static map | Sleep, diet families, hidden-inputs modules · sectioned questionnaire · static systems map | Internal users can complete questionnaire and see their static map |

| 2 — Algorithmic mirror MVP | Mortality model · single trajectory view · sensitivity sliders · methodology page | Trajectory view live, methodology public, opt-in life expectancy live |

| 3 — Full systems map + remaining modules + lab upload | All 8 sub-environment modules · modality library · life-stage modules · optional lab PDF upload | Library complete · lab upload working |

| 4 — Video library + family mode + clinician export | Interview videos · family-mode parent-anchored child view · PDF export for clinicians | Video library live with first 12 interviews · family mode tested with parent users |

| 5 — Self-hosting + federation | Self-hosting option for advanced users · federation with other AF platforms (Whiteboard bridge, possibly Halo for secure messaging with clinicians) | Self-host docs published · bridge to Whiteboard live |

16. Open questions

1. Final name. Biome is the working lead; check availability across .com/.net/.org and trademark before lock.

2. Funding model. No commercial entanglement = no ad / affiliate revenue. Options: nonprofit grant (likely fit for AF), donor-funded, optional user subscription with anti-influence governance, foundation partnership. Decide before Phase 1.

3. Lab upload. PDF parse via PII-stripped local processing? Phase 3 likely. Heavy data category.

4. Clinical instruments. PSS-10 yes. PHQ-9 / GAD-7 — out of v1 unless paired with clinical partnership.

5. Whiteboard bridge. Shared user account? Cross-link from Whiteboard's life-skills modules? Probably yes, in Phase 5.

6. Family mode in v1? Lean: no. Add in Phase 4. v1 is adults only.

7. Children's content. Out of scope for v1. Possibly Phase 4+ under family mode, parent-anchored. Never independent child profiles. Aligns with portfolio child-safety posture.

8. Editorial board / source review. Volunteer scientific advisory board with rotating membership? Worth exploring in Phase 0.

17. Reference: portfolio integration

| Adjacent platform | Relationship |

|---|---|

| Whiteboard | Bridge in Phase 5. Whiteboard's life-skills + continued-learning modules cross-link to Biome modules. Biome users in Whiteboard's adult tier get cross-app continuity. |

| Halo | Encryption patterns reused. Possible Phase 5 use: secure clinician messaging. |

| Conduit | When Conduit is live, Biome's third-party data integrations (AirNow, EWG, PubMed) routed through Conduit for managed-key vaulting and short-lived credentials. |

| Family Detective Lab | Distinct audience and purpose. No direct integration. |

| TruthMark / ShieldScan / Exodus | Brand-posture cousins (transparency, no commercial entanglement, citation-first). Shared visual identity through editorial series. |

End of v0.1.

Next: name lock → Phase 0 kickoff.

← Back home