Confidential · April 2026

Zero to Hero
Scaling Blueprint

Product capability, hardware transition, operating audit, unit economics, and release gates. Voice/audio remains protected while every visible product surface scales.

Product Max Stack

Every Layer Must Earn Trust

The live website now behaves like the control plane for the future toy: the child gets play, learning, and emotional support; the parent gets clarity and control; the team gets measurable release gates.

Kid Delight Layer

Adaptive persona gamesStory arcs across sessionsMovement breaksCreative missionsEmotion naming

Parent Assurance Layer

Plain-language summariesConcern escalationConsent logsTopic controlsProgress exports

Learning Growth Layer

Vocabulary growthSTEM pathwaysFolklore packsDifficulty tuningMilestone maps

Hardware Bridge Layer

Device pairingOTA updatesBattery statesFactory QAOffline fallback copy

Operating Control Plane

One Gate System Across Web, Toy, and Release

The readiness model keeps every scale decision tied to an owner, threshold, evidence source, and workflow handoff. No new claim ships without a gate and an artifact.

Current readiness

80

Build gates intentionally score below locked gates until hardware pairing, OTA, factory QA, and release cadence have production evidence.

Voice/audio regression guard

locked

Owner: Voice

No implementation diff in protected TTS/STT/voice-engine paths

Age-banded talking and data policy

locked

Owner: Product

4-5 parent-led only; 6-12 core toy range; teen mode is study/coaching, not companion framing

Visible product surface coverage

watch

Owner: Release

Homepage, chat, parent, devices, quality, scale, and deck routes present

Hardware bridge readiness

build

Owner: Device

Pairing, scoped tokens, OTA, factory QA, and offline states defined before toy pilot

Parent trust and auditability

watch

Owner: Trust

Safety alerts, transcripts, quality scores, settings, and device status are discoverable

Operating loop and release cadence

build

Owner: Ops

Every scale release records checks, owners, thresholds, and next action in one artifact

Child session to parent insight

Child gets a short age-fit response; parent gets a useful summary signal.

Chat -> transcript -> quality score -> parent next action

Web account to toy endpoint

Toy behaves as a controlled endpoint of the parent account, not a separate companion system.

Parent profile -> device pairing -> scoped token -> OTA/factory QA

Release to live deployment

Ship only when product, safety, voice, and route gates are recorded.

Audit -> lint -> build -> push -> Vercel preview -> smoke

Zero-Tolerance Audit

Release Gates Before Scale Spend

Voice quality is guarded by the existing TTS probe. The broader stack gets a separate readiness audit so UI routes, hardware gates, safety posture, and operational ownership are checked without touching audio code.

Maximum Threshold Checks

MetricTargetOwner
Homepage, chat, parent, scale, deck route availability100% HTTP 200Release
Parent modules linked from dashboard/sidebarNo dead visible surfaceProduct
TTS first audio byteUse existing probe; no voice regression allowedVoice
Child-safe response checks0 critical guardrail misses in smoke packSafety
Hardware pairing happy path< 90 seconds parent to active toyDevice
Production incident rollback< 10 minutes to previous deployOps

Run product audit

npm run audit:scale

Run voice audit

scripts/probe-prod.ts

Run route smoke

scripts/smoke-prod.sh

Hardware Toy Integration

Web First, Toy Next, Same Safety Core

The toy should be a trusted endpoint for the same parent-controlled system, not a separate product with a separate safety model.

Non-negotiable principle

Parent permissions, safety boundaries, learning profile, and device status must be visible from the dashboard before any toy-only feature ships.

1

Web account becomes source of truth for child profile, permissions, personas, language, and safety policy.

2

Toy firmware receives only scoped session tokens and never stores raw long-term child conversation history.

3

Every device must pass mic, speaker, button, battery, Wi-Fi, LED/haptic, enclosure, and thermal checks before pack-out.

4

OTA updates use signed firmware, staged rollout rings, rollback, and parent-visible release state.

5

Offline states are explicit: wake, no network, low battery, privacy muted, update required, and factory reset.

Age-Graded Intelligence

Talking Level, Data Level, Safety Level

The product starts from a parent-led 4-5 mode, then scales into the strongest 6-12 hardware range. Teen support is framed as study and coaching, never as an AI companion.

Ages 4-5

Sprout

Parent-led web sessions only; no autonomous toy companion mode.

Session cap

8 min

Talk level

18-35 words. Concrete, playful, one-step prompts with parent handoff.

What kids get

songs and rhymes, colors and counting, simple feelings, pretend play, daily routines

What parents get

session length, mood cues, new words heard, parent co-play prompts

Ages 6-8

Explorer

Core toy/web mode with short supervised sessions.

Session cap

12 min

Talk level

25-55 words. Simple why/how reasoning, turn-taking games, and tiny learning quests.

What kids get

stories, riddles, phonics, number sense, movement breaks

What parents get

topic interests, vocabulary growth, confidence signals, safety alerts, next activity

Ages 9-12

Builder

Core advanced mode for projects, homework scaffolding, and creativity.

Session cap

18 min

Talk level

40-85 words. Multi-step reasoning with examples, reflection, and child-led choices.

What kids get

STEM projects, creative writing, debate practice, current affairs for kids, study habits

What parents get

skill mastery, question patterns, learning gaps, project progress, peer/social flags

Ages 13-18

Mentor

Supervised study and coaching mode; not positioned as an AI friend.

Session cap

25 min

Talk level

60-120 words. Structured coaching, source-aware learning, planning, and reflection.

What kids get

study planning, exam prep, career exploration, AI literacy, well-being check-ins

What parents get

goals, study plan adherence, risk flags, strengths, parent-visible summaries

Phase 0 — Current State (Software-Only Demo)

What We Have Today

Live web demo at neev.truss.biz. Voice pipeline working end-to-end. No hardware costs yet — pure software validation.

Current Monthly Software Costs (Live Demo)

ServiceProviderCostNotes
LLM (Chat AI)Azure OpenAI gpt-4.1-miniUsage-basedNVIDIA DeepSeek remains available as fallback
STT (Speech-to-Text)Azure SpeechUsage-basedHindi + Indian English recognition
TTS (Text-to-Speech)Azure Speech Neural TTSUsage-based48 kHz MP3 output with persona SSML
HostingVercel (Hobby → Pro)$0–$20/moNext.js serverless
Total (100 daily users)~₹2,000–4,000/mo~$25–50/mo all-in

Layer 1 — Hardware BOM (Bill of Materials)

Electronics Cost Per Unit

ESP32-S3 based voice module. Prices from LCSC, DigiKey, and Alibaba at volume tiers. All in INR at ₹84/USD.

Component Breakdown

Component100 units1,000 units5,000 units10,000 units
ESP32-S3-WROOM-1-N16R8₹327₹210–252₹168–210₹151–185
INMP441 MEMS Microphone₹126–210₹42–67₹34–50₹29–42
MAX98357A Speaker Driver₹151–252₹50–84₹42–67₹34–50
Speaker (28mm, 1W)₹25–67₹17–34₹13–25₹13–21
LiPo Battery (1200mAh)₹126–210₹67–101₹50–76₹42–59
USB-C + Passives + Antenna₹42–84₹25–42₹21–34₹17–29
PCB Fabrication + Assembly₹300–800₹80–200₹40–120₹30–80
Total Electronics BOM₹1,097–1,950₹491–780₹368–582₹316–466

Layer 2 — Plush Toy + Packaging

Soft Toy Manufacturing

Baby-grade hypoallergenic fabric. Custom moulds. Premium rigid box packaging with embossing. Indian manufacturers (Kuddl Toys, IndiaMART suppliers).

Manufacturing + Packaging

Component100 units1,000 units5,000 units10,000 units
Plush body (hypoallergenic, custom)₹400–700₹200–400₹150–300₹120–250
Packaging (premium rigid box)₹200–350₹85–150₹42–85₹21–50
QC + Final Assembly₹100–200₹50–100₹30–60₹20–40
Total Physical₹700–1,250₹335–650₹222–445₹161–340

Layer 4 — Software COGS Per Device Per Year

AI & Cloud Costs at Scale

Per-device annual costs assuming 5 conversations/day, 10 turns each. Current launch stack: Azure OpenAI for chat and Azure Speech for STT/TTS.

Annual Software Cost Per Active Device

ServicePer InteractionDaily (5 conv)MonthlyAnnual
Azure OpenAI gpt-4.1-miniUsage-basedMeasured in pilotMeasured in pilotTBD
Azure Speech STTUsage-basedMeasured in pilotMeasured in pilotTBD
Azure Speech Neural TTSUsage-basedMeasured in pilotMeasured in pilotTBD
Cloud hosting (amortised)——₹30–60₹360–720
Total per device/yearTo be locked after Azure pilot telemetry

Critical insight: Voice cost is the primary software COGS lever. Launch telemetry should track Azure Speech STT seconds, TTS characters, cache hit rate, and average spoken-response length before committing public per-device economics.

Voice Cost Optimisation Scenarios

ScenarioAnnual/DeviceMonthly/Device
Azure launch stackMeasured during pilotDepends on usage
Higher cache hit rateLower TTS spendPre-cache common prompts
Shorter spoken-response budgetLower chat + TTS spendBest for child attention
TargetLock after live telemetryPublish after pilot

All-In Unit Economics

Cost vs Price — The Margin Stack

Unit Economics at 1,000 Scale (Optimised TTS)

Line ItemCost (₹)% of Revenue
Electronics BOM₹6007.5%
Plush + Packaging + QC₹5006.3%
BIS amortised (1,000 units)₹1301.6%
Hardware COGS₹1,23015.4%
Shipping (D2C avg, prepaid)₹1001.3%
GST (18% on ₹8,000)₹1,22015.3%
Landed Cost₹2,55031.9%

Retail Price

₹8,000

incl. GST

Hardware Gross Profit

₹5,450

per unit sold

Gross Margin

68%

before CAC & opex

Revenue Model — Subscription Tiers

Customer Pricing vs Actual COGS

Three tiers. Free drives adoption. Pro is the engine. Max is high-margin premium.

Subscription Unit Economics (Optimised TTS Stack)

TierPriceSoftware COGS/moGross MarginAnnual Revenue
Free₹0₹50–80-100%₹0
Pro₹799/mo₹250–29064–69%₹9,588
Max₹7,999/mo₹250–290 + ₹2,000 expert71–72%₹95,988

Lifetime Value (LTV) Per Household

MetricFree → Pro ConvertDirect ProMax
Device revenue₹8,000₹8,000₹8,000
Year 1 subscription₹6,393 (8 mo)₹9,588₹95,988
Year 2 subscription (70% retention)₹6,712₹6,712₹67,192
2-Year LTV₹21,105₹24,300₹1,71,180

Distribution Costs

Channel Economics

Channel Margin Comparison

ChannelCommissionOther FeesEffective Take RateNet to Neev (₹8,000)
D2C (own website)0%Payment gateway 2%2%₹7,840
Amazon.in0% (<₹1K) / 8–15%₹30 closing + shipping~12–18%₹6,560–7,040
Flipkart7–12%2% collection + shipping~15–22%₹6,240–6,800
FirstCry15–25%Listing fees~20–28%₹5,760–6,400
Croma / Hamleys30–40%Slotting + display~35–45%₹4,400–5,200
B2B SchoolsBulk ₹1,499/unit—81% discount₹1,499

Strategy: D2C first (highest margin). Marketplace for discovery. Retail only after brand is established. B2B schools at cost for market penetration and word-of-mouth.

Phase-Wise Scaling Plan

From 100 to 100,000 Units

Phase 1

Beta & Validate

Month 0–6 · 100 units

₹1 Cr (F&F Round)
Hardware (100 units @ ₹2,800)₹2.8L
BIS certification + legal₹2.5L
Cloud + API costs (6 mo)₹0.3L
Team (CTO + Content Lead, 6 mo)₹3.0L
Trademark + IP₹0.6L
Contingency₹0.8L
Total spend₹10.0L
Projected revenue₹0 (free beta)
Phase 2

D2C India Launch

Month 6–12 · 1,000 units

₹1 Cr (Angel Round)
Hardware (1,000 units @ ₹1,600)₹16.0L
Marketplace setup (Amazon/Flipkart/FirstCry)₹1.5L
D2C website + payment integration₹1.0L
Marketing (WhatsApp communities, PR)₹3.0L
Cloud infra scale-up₹1.0L
School B2B pilot (50 units @ cost)₹0.8L
Team expansion + ops₹6.0L
Total spend₹29.3L
Projected revenue₹80L (1,000 × ₹8,000)
Phase 3

Scale & Expand

Month 12–24 · 5,000–10,000 units

₹3.5 Cr (VC Round)
Hardware (5,000 units @ ₹900)₹45.0L
Regional language content (Tamil, Telugu, Bengali)₹5.0L
Retail entry (Croma, Hamleys)₹8.0L
Marketing + brand campaign₹15.0L
Cloud + API at scale (2,000+ Pro subs)₹6.0L
Full team (10 people, 12 mo)₹36.0L
South Asia + NRI expansion₹5.0L
Series A prep + legal₹3.0L
Total spend₹1.23 Cr
Projected revenue₹4.0–8.0 Cr (devices + subscriptions)

Competitive Pricing Landscape

Where Neev Sits

Market Pricing Comparison (April 2026)

ProductHardware PriceSubscriptionIndian LanguagesScreen-Free
Neev₹8,000₹799/mo ProHindi + EN live · 7 on roadmapYes
Miko Mini₹8,999Miko Max (optional)Partial HindiNo (screen)
Miko 3₹19,490₹37,999 w/ 2yr MaxPartial HindiNo (screen)
BubblePal$89 (~₹7,500)$40/yrNoYes
Bondu$199 (~₹16,800)TBANoYes
Amazon Echo Dot Kids₹4,499Free (Alexa)Basic HindiYes

Price positioning: Neev at ₹8,000 is cheaper than BubblePal ($89), half the price of Bondu ($199), and dramatically cheaper than Miko 3 (₹19,490). The only cheaper alternative is Echo Dot Kids at ₹4,499 — but it has no child safety architecture, no Indian language depth, no parental dashboard, and no curriculum alignment.

The Bottom Line

Year 1 Projections

1,000

Units Sold

₹80L

Device Revenue

₹9.6 Cr

Pro ARR (at 1K subs)

68%

Hardware Margin

Year 1 LTV per household (device + 12 months Pro): ₹17,588. CAC payback target: under 6 months. Series A trigger: 5,000+ units, 2,000+ Pro subscribers, NPS 65+.

Neev Scaling Blueprint · Truss Studios · April 2026 · Confidential