Canonical file: docs/BRIEF.md — this page renders it verbatim. Relative links may point at repo files rather than lab pages.

Vocal ER — Product Brief

Source: Nina × Ian planning conversation, 2026-08-03 (full transcript). This brief is the distilled, working version — update it as decisions land.

One-liner

Voice rehabilitation for everyone who depends on their voice — a real voice coach's recovery method as an app, with an AI triage nurse that routes you to the right exercises, growing into a platform where speech therapists and ENTs treat their own patients.

The people

Nina — voice coach. Brings the method, the exercise content, the credibility, and the clients (students signed with Atlantic/Sony, touring singers). Teaches 100% virtually for 6 years — so her method already works over consumer mics. Has ~1–1.5 years of recorded lessons + written notes (stored locally only; being backed up + shared via Google Drive). Has an existing very-basic build (Replit + Claude, on GitHub) and a website — links pending. Not a certified SLP — which is why positioning stays wellness/education-side and certified professionals get a formal place in the product (see booking + Tier 2).
Ian (ICG Labs) — product + engineering.
Barry — Nina's friend, works in HIPAA compliance professionally. Advised HIPAA-from-day-one; available to review code/architecture.

Product strategy — two tiers, one platform

Consumer (launch tier): educational + rehabilitative exercises for anyone with voice issues. Browse & self-select, or take the AI triage. Content built and curated by Nina.
Medical professional (growth tier): SLPs and ENT/laryngologists onboard as providers — upload their own content, manage their own patients. The patient experience ≈ the consumer experience; the professional layer is an add-on on the same infrastructure, not a rebuild. Nina is the first "pathologist entry" demonstrating the model.
Long-term vision: break the silos — cross-disciplinary education between voice coaches, speech therapists, and laryngologists, who today barely talk to each other.

Core features

Feature	Notes	Tier
Exercise library	Browse + self-select (this is roughly what Nina's current build does)	v1
AI triage nurse	Voice/phone-call style preferred over text chat. AI version of Nina asks intake questions ("How's your voice feeling? What do you use your voice for?"), routes via decision tree, auto-assigns ~5–6 exercises; periodic check-ins reassess and reassign. Users self-report symptoms — the AI does not "listen and diagnose" yet. Early version: audio + cartoon/graphic visual with chat bubbles ("like an interactive video game"). Strict on-topic guardrails. Skippable — experienced users go straight to the library.	v1–v2
Legal disclaimer	Not a substitute for professional care — everywhere it matters	v1
Direct booking	Book Nina at her coaching rates for users who want a real person — plus a bench of recommended certified pros (voice coaches, SLPs, doctors) for scale and for anything beyond coaching scope	v1–v2
Practice video uploads	Users record themselves doing exercises, upload for review	v2
In-app video chat	Live virtual sessions through the app	v2 (PHI)
Laryngoscopy video uploads	Patients upload scope videos (HIPAA-protected); potential anonymized research library over time	Tier 2 (PHI)
Provider content builder	Drag-and-drop so professionals reuse existing exercises instead of building from scratch	Tier 2

Content & user categorization (two layers)

Symptom-based (core): the voice recovers the same way regardless of how it was injured — like a sprained wrist heals the same for anyone. Core exercises apply broadly.
User-type (additive): niche channels layer extra guidance on top — e.g. touring singers get dietary, sleep, and show-timing guidance in addition to standard exercises.
Launch channels: general voice recovery + touring singers. Later niches (public speakers, autistic nonverbal children, …) built out by the relevant professionals on Tier 2.

Pricing & membership

Anchor: stay under $175/month — the cost of a single speech-therapy session.
Tiered model discussed: basic (exercises only) → premium (AI advice) → premier (community access for niches like touring singers).
Decided in-meeting: tiers/paywall don't need to go into the first round — build first, monetize after. (The debate — paid-from-start signals professionalism and commitment vs. free launch for feedback — stays open for later.)
North-star reference product: the home-exercise app Ian's physical therapist used, insurance-covered — Nina: "this is what I want to do." Research identified that category as Physitrack/PhysiApp (see RESEARCH.md) — the Tier 2 template.
Long-term revenue goal: insurance reimbursement ("when insurance pays for things, they pay a lot") — requires HIPAA compliance and shapes Tier 2 design (see RESEARCH.md on RTM billing codes).

Technical direction (decided in the meeting)

Web-first — faster iteration, no app-store friction, works across devices. Mic + camera access in the browser is sufficient for the core use case.
Web can't do push notifications or depth sensors → reminders via SMS and/or email, user-selectable.
Mobile later: wrap the Next.js app with Capacitor for iOS/Android with over-the-air updates.
HIPAA-compliant architecture from the start (Barry's advice: retrofit = rebuild). Key implications: secure private infrastructure, strict limits on what even the app owners can see (e.g. diagnoses), audit trails. Note: the ICG Labs default (shared multi-tenant Supabase) is not appropriate for the PHI tier — stack decision pending research.
Claude can assist with HIPAA compliance checks; outputs double-checked, and Barry reviews.

AI voice analysis (future vision, groundwork now)

The dream: an AI voice coach that listens and diagnoses (e.g. resonance placement). Not feasible at coach level today, but the infrastructure is advancing fast.
Groundwork starting now: back up Nina's ~1–1.5 years of lesson recordings, categorize and analyze them (audio/video + her notes) so the dataset exists when the tech does.
The decision-tree triage bot and the future listening coach develop in parallel — the triage bot doesn't wait.

Brand

Name: Vocal ER (working; per Ian 2026-08-03). Evokes "vocal emergency room."
Current look: red/blue ambulance colors; a nurse mascot as attention-getter (referenced as "Blink-182 nurse" style — ⚠️ must be original artwork, inspired-by only; the actual Enema of the State nurse imagery is copyrighted).

Open questions

Final name/brand (Vocal ER working title; check trademark + domain availability — there is no .er TLD, so vocal.er is out)
Pricing amounts + tier structure (paywall explicitly deferred from round one)
Feasibility/appetite for an AI video replica of Nina (early version = audio + cartoon visual)
IP ownership of provider-uploaded content (Tier 2)
HIPAA-aware stack choice (pending RESEARCH.md) and where exactly the PHI boundary sits in Tier 1
Assessment of Nina's existing Replit build — reuse anything, or greenfield?
Public build-in-public demo vs. internal-only until launch (health app + partner content = default private for now)
Nurse mascot artwork (original commission)

Action items from the founding meeting

Nina: upload recorded lesson videos to Google Drive + share with Ian · share GitHub repo + website link · back up all local files to cloud · research HIPAA requirements with Barry · attend the November speech-therapist conference (networking) · start recording more lessons with student permission for the AI dataset · consider hardware upgrade (MacBook Pro / Mac Mini) — hold until needs confirmed.

Ian: send Nina the summary text with everything needed to get started (draft here) · set up this project foundation ✅ · research the tool landscape ✅ · prototype the recording pipeline (scan/categorize her videos, pull metadata, per-video voice analysis — "messy and experimental") · build the decision-tree triage chatbot with her.