Lumbiota · Product Specification · v1

How the app works

Prepared for Dr. Gabriela Sheets. This spec translates the adoption framework into screens, flows, and a data model — the scope needed to build and demo Lumbiota's V1.

1Adoption thesis, restated

Lumbiota attaches to behavior a family already has — planning meals, cooking dinner, feeding a household — and makes each moment slightly better. The phone is the whole product. Three rules govern every decision:

  1. Value before data. A protocol-aligned recipe arrives before any microbiome onboarding.
  2. Science is the engine, never the interface. The user sees dinner, not diagnostics.
  3. Hardware is optional, never required. Casting delights; a single iPhone suffices.

2Who it's for

Life stageWhat they need from the app
ExpectingTrimester-aware nutrition that quietly builds maternal microbial diversity ahead of birth.
New parent, infant 0–12 moAllergen-introduction guidance and purée-stage recipes timed to pediatric protocol windows.
Household running multiple protocolsOne shared meal that serves two or more tracks at once, without cooking twice.

3The core loop

PlanCookTrackLearn

Home surfaces one action (Plan). Cook Mode executes it (Cook). Protocol Alignment updates in response (Track). The First 1,000 Days timeline contextualizes why it mattered (Learn). No step requires leaving the kitchen.

4Information architecture

ScreenPurpose
OnboardingLife-stage picker → 3 recipes, grounded in developmental microbiome science. No account required.
HomeProtocol Alignment (hero metric) + tonight's recipe + journey preview.
First 1,000 DaysHorizontal journey from conception to toddler; current position, active protocol, what's next.
Recipe detailIngredients, swap options, protocol rationale, "start cooking."
Cook ModeOne step per screen, embedded timers, cast entry point.
Family TableMulti-protocol households: swipeable or split-view tracks, live ingredient-swap recalculation.
Depth / DiagnosticsDiversity scores, taxa breakdowns, longitudinal trends — one tap below Home.
Cast pickerAirPlay / Google Cast / StandBy — surfaced from the share icon, not a tutorial.

5Key flows

Flow A

First two minutes

  1. Download, open app.
  2. Pick a life stage (Expecting / Infant / Toddler / Household).
  3. Receive three protocol-aligned recipes immediately.

No account wall — the first three recipes need nothing but a life stage. Sequencing-based personalization is available immediately for families who already have a practitioner's result, and encouraged once the app has proven itself.

Flow B

First cook → cast discovery

  1. User taps a recipe, enters Cook Mode.
  2. Calm, legible, one-step-at-a-time experience — the "this is different" moment.
  3. User taps the share icon mid-cook, discovers AirPlay / Cast / StandBy on their own.
Flow C

Deep data unlock

  1. After repeated use, Home surfaces an optional prompt to connect sequencing data — never blocking.
  2. User connects an existing sequencing result from a qualified practitioner, or is guided toward one.
  3. Diversity scores and taxa trends unlock under Depth; Home's hero metric sharpens in accuracy.
Flow D

Family Table cooking

  1. A household has two active protocols (e.g. maternal high-fiber + infant allergen-intro).
  2. Cook Mode presents both tracks — split view when cast to a wide screen, swipeable on phone.
  3. User taps an ingredient swap ("out of chicory root"); fiber-to-polyphenol equivalence recalculates instantly.

6System architecture

Phone-first, cast-optional. Cook Mode is built once as a single responsive layout; casting renders that same view on a remote screen. There is no voice-skill certification and no smart-speaker platform risk.

Lumbiota iOS App
State machine · profiles · diagnostics · recipes · Cook Mode
Cast targets
optional, one tap, none required
AirPlayGoogle CastStandBy / iPad
Voice control (Siri Shortcuts) ships later, only if usage data asks for it.

7Conceptual data model

EntityKey attributesRelates to
Protocollife stage, active window, target nutrientsHousehold, Milestone
Life Stagetrimester / infant month / toddlerProtocol
Recipeingredients, prep time, protocol tags, swap mapProtocol, Track
Householdmembers, active TracksTrack, Protocol
Trackmember, current Protocol, texture/stageHousehold, Recipe
Diagnostic Resultdiversity score, taxa breakdown, dateProtocol, Milestone
Milestonestage marker, status (past/now/upcoming)Protocol
Cast Sessiontarget type, active recipe, step indexRecipe

8V1 scope vs. deferred

In V1Deferred
Hero metric, First 1,000 Days timeline, recipe engineSmart-appliance APIs (multi-cookers, ovens)
Cook Mode + AirPlay / Cast / StandByVoice control (Siri Shortcuts)
Family Table, ingredient-swap recalculationFull longitudinal diagnostics dashboard

Why defer smart appliances: manual temperature check-ins plus Live Activity timers cover ~95% of the value at 5% of the engineering cost. Appliance APIs get built when demand justifies the maintenance burden.

9Practitioner network & licensing PHASE 2

Dr. Sheets' practice today is bottlenecked on her own hours: sequence a client's microbiome, interpret it, build a custom plan, then manually stay in the loop between sessions. Her own site already lists "Healthcare Practitioners" as a service line — consultations, lectures, training other clinicians. The app is the natural product wrapper for that: it turns a one-to-one consulting practice into a licensable one, run from anywhere.

What changes for her

The app absorbs the daily-touch work — recipes, Cook Mode, Family Table, Protocol Alignment tracking — so clients get value between sessions without her being physically present. Her role shifts from "delivers the daily plan" to "reviews sequencing data and adjusts protocols remotely." That's a job doable from a laptop anywhere, with occasional video check-ins standing in for a physical practice.

Practitioner Console — the new surface this requires

Everything specified above (sections 1–8) is consumer-facing: a family managing its own protocols. None of it lets a practitioner manage a caseload. That's a distinct surface, built on the same design tokens, not a repurposing of existing screens.

ScreenPurpose
Caseload rosterEvery active client/household at a glance, sorted by who needs attention (alignment drop, unread sequencing result, protocol due for review).
Sequencing intakeUpload or connect a client's microbiome sequencing result; the practitioner reads and annotates it directly — this is where her actual diagnostic expertise lives in the product.
Protocol assignmentTurn a sequencing read into an active Protocol for that client's Track — the same Protocol entity the consumer app already runs on (see §7).
Cross-client trendsAlignment and adherence across the whole caseload, not just one household — spot who's drifting before they churn.
Async check-inLightweight messaging tied to a specific Protocol or Milestone, replacing ad-hoc email/text between sessions.

Licensing model

TierWho it's forRevenue
Sheets-run instanceHer own direct clientsExisting consulting fee, now with a lower cost-to-serve per client
Certified Practitioner licenseMidwives, doulas, integrative nutritionists she trains and certifies on her methodologyPer-seat or per-active-client license fee, or a revenue share on each practitioner's client billing

Why licensing beats training alone: she already trains practitioners informally (lectures, consultations). Wrapping that training in a licensed app means every certified practitioner runs on her actual protocols — not their memory of a workshop — so quality holds as the network grows, and she earns recurring revenue from a caseload she never personally touches.

What clients get out of it

The same daily companion a solo family gets — a recipe, a cook, a journey — but now backed by a practitioner who reviews their real sequencing data periodically instead of guessing from symptoms. Fewer "what do I eat tonight" messages land in the practitioner's inbox, because the app already answered it.

This is explicitly out of V1 scope (§8) — it depends on the consumer app proving the adoption thesis first. Sequenced correctly: consumer V1 → Deep Data unlock (§5, Flow C) validates that families actually connect sequencing results → Practitioner Console productizes the review side of that same data.

10Success metrics, by funnel stage

StageWhat proves it worked
Download → value% of new users who view a recipe within 2 minutes, no account created
First cook → hook% who complete a full Cook Mode session; % who discover casting unprompted
Deep data → commitment% who connect a sequencing result after ≥3 cooks (opt-in rate, not prompt-driven)
Family Table → moatHousehold retention at 90 days vs. single-profile retention
Practitioner networkCertified practitioners active at 6 months; average caseload size per practitioner without quality drop-off