The Missing Middle
Consumers have no trusted product that connects what the body recorded, what the mind noticed, and what still feels present in the morning. Wearables deliver metrics. Reflection apps store words. Almost nothing turns the two into personal context a companion can carry forward.
| Category | Passive Trackers (Oura, Apple Watch) |
Reflection Apps (journals, meditation) |
Somnora |
|---|---|---|---|
| Body Context | Strong (biometrics) | Minimal | HealthKit trends vs your baseline |
| Inner Context | None | Self-reported only | Dreams, daily check-ins, ideas |
| Relationship Layer | None | Static content | Nora: memory, initiative, continuity |
| User Outcome | Metrics without meaning | Reflection without context | Context that compounds overnight |
The Reflective Pipeline
Somnora combines wearable signals, private reflection, and conversation so users understand what they are carrying, then decide what deserves attention next.
Phase 1
Capture
Voice or text in seconds: dream, daily check-in, or Eureka fragment. Watch capture wakes the phone conversation.
Phase 2
Context
HealthKit sleep timing, HRV, and resting heart rate read against the user's own rolling baseline.
Phase 3
Conversation
Nora responds as a companion: short-first, observant, register-mirroring, never a score or diagnosis.
Phase 4
Patterning
On-device motif detection, nightly memory consolidation, and at most one earned initiative per day.
Outcome
Clarity
A private archive of what repeats, plus a calmer next step when the evidence has earned it.
A Human System, Technically Grounded
Dreams are the wedge, not the ceiling.
Dreams create an emotionally resonant entry habit. From there Somnora expands into daily check-ins, Eureka idea capture, Mindful wind-down tools, and Dreamcatcher art. The system is designed to feel calm to the user and rigorous underneath.
01. Body
Sleep & Autonomic Trends
Apple Watch and HealthKit provide timing, stages, HRV, and resting heart rate. Trends steer Nora's tone without turning into a leaderboard.
02. Capture
Dream, Daily, Eureka
One companion across morning dreams, daytime residue, and creative fragments. Capture is conversational, not a form ceremony.
03. Nora
The Relationship Layer
Nightly on-device synthesis can start the conversation. Cross-night memory, motif echoes, and earned check-ins make return visits feel personal.
04. Continuity
Art, Patterns, Wind-Down
Dreamcatcher turns a dream into polaroid-style art. Mindful offers evidence-informed wind-down tools. Patterns stay private and user-owned.
The Experience
A low-friction loop built for repetition: easy enough to start in seconds, valuable enough to revisit over nights and days.
-
01
10-Second Start
Voice note, typed fragment, or Watch capture when something lingers. -
02
Ambient Context
HealthKit layers sleep timing and autonomic trends quietly in the background. -
03
Clearer Direction
Nora reflects what repeats and helps decide what deserves attention next, including a concrete next step when it is earned.
Live product UI · Dream Journal and Eureka
Efficient Architecture, Built for Trust
Heavy understanding runs on device. Cloud calls go through Somnora's own proxy with privacy sanitization. Richer model work is reserved for moments that need a companion response or dream art.
Private Reflection &
Health Context
On-Device Memory
Motifs, embeddings, nightly consolidation, autonomic trends, initiative gating.
Cloud Run Proxy
Vertex for Nora chat · Imagen for Dreamcatcher art
Production path: app → somnora-proxy → Vertex AI
Early Signals, Clear Pull
Shipped Consumer Product
Somnora is live on the App Store as a native iPhone and Apple Watch product. Core loops for Dream, Daily, Eureka, Mindful, Insights, and Dreamcatcher are real, gated honestly, and privacy-hardened.
Capital-Efficient Build
Solo-founded and end-to-end owned: iOS client, Watch companion, Cloud Run proxy, and Dream Painter service. Cost discipline is designed in through quotas, on-device first passes, and selective model routing.
Expansion Beyond the Wedge
Product direction already extends from dreams into daily reflection, idea development, wind-down support, and recurring pattern review, widening the habit surface without abandoning the original entry.
Defining the Middle Layer
Metrics
Reflection
Why Now
Wearables are mainstream, on-device AI can finally feel companion-like, and privacy has become a product decision consumers actively make.
Target Unit Economics
Subscription product, $9.99/mo premium tier (Founders yearly available at launch).
Premium lifts Dreamcatcher art capacity, Nora chat allowance, and deeper pattern reads.
Comparing monthly COGS stack vs. premium revenue distribution.
Modeled with 15% platform fee assumption.
Free User (Modeled)
Includes capped Nora chat, limited Dreamcatcher renders, and full privacy controls. COGS stay low through weekly quotas, caching, and on-device first passes.
Premium User @ $9.99
Unlocks higher Dreamcatcher and Nora allowances plus deeper reads. Architected to support a target gross margin >65% at scale.
Privacy as the Trust Moat
When a product touches intimate material, trust is not a policy page. It is the product. Privacy, restraint, and user control are part of the experience itself.
-
✓
Architecture:
Motifs, embeddings, autonomic trends, and nightly synthesis run on device. Cloud chat is sanitized before egress through Somnora's proxy. -
✓
Model Training:
We do not train public models on user content, and we do not sell user data. -
✓
User Control:
Local export, deletion, Face ID app lock, editable memory, and clear ownership boundaries are built in from the start.
Leadership
-
→
The Storyteller:
12 years in documentary cinematography, helping shape emotionally precise stories and projects that raised $4M+ for conservation. -
→
The Builder:
Self-taught iOS and Python engineer who built Somnora end to end, then created Manifold—an open-source Lambda GPU control plane—to sculpt and tune Nora from its initial Gemini harness. -
→
The Origin:
A long personal habit of capturing vivid dreams, late-night ideas, and daily residue became the first wedge. LLMs made it possible to connect that reflection with wearable context in a way that finally felt useful.