Investor Brief

Overview

36 million US dogs suffer from noise anxiety. The gold-standard solution costs $2,100 per 10-session course and fewer than 80 board-certified specialists practice in the entire US. Every connected-pet hardware company doubled down on remote-engagement features in 2025, leaving the therapy use-case entirely uncontested. Pawvlov is the first device-grade implementation of a clinically grounded behavioral protocol — autonomous, body-language-timed, and silent enough that an anxious dog doesn't startle. § Sourced

Market Research

US pet anxiety category, 2026. Bottoms-up SAM modeled from anxiety-diagnosed dogs in storm-belt and apartment-firework regions at $60K+ HHI.

Target Segments

• Storm-belt urban + suburban dog owners (FL / TX / Southeast): ~12M US households △ Modeled
• Apartment dog owners exposed to fireworks (July 4 + NYE): ~8M US households △ Modeled
• Premium pet-wellness spenders (BarkBox / Chewy autoship / pet insurance): ~6M US households △ Modeled

Key Findings

• 36M US dogs (≈1 in 3 household dogs) suffer from noise anxiety. § Sourced
• Certified veterinary behaviorist protocol runs ~$2,100 per dog (10 sessions × $200–250). § Sourced
• Furbo / Petcube treat-dispenser noise is iatrogenic for anxious dogs — Reddit threads document anxious dogs avoiding the room with the dispenser. ❝ Customer
• 10K units/year @ $249 + 10K subs @ $120/year = $3.7M ARR steady-state year 2. △ Modeled
• ESP32-S3 onboard inference now runs a 4-class dog-pose classifier at 5 fps with sub-second latency for under $4 SoC cost. ◌ Assumption

Competitive Landscape

Pawvlov is the only product delivering veterinary-behaviorist-grade systematic desensitization autonomously, at 3am, in the room where the anxiety actually compounds. Furbo and Petcube doubled down on remote-engagement features in 2025, leaving the therapy use-case entirely uncontested. The ceiling isn't the $249 connected camera — it's the $2,100 DACVB consult Pawvlov undercuts by 90%.

Structural differentiators

1. Closed-loop autonomous protocol — only device that reads dog state AND delivers reinforcement timed to calm. △ Modeled
2. Therapy-grade dispenser engineering — sub-34 dB auger is a deliberate inversion of the ~65 dB Furbo click. △ Modeled
3. Trigger-actuated, not owner-actuated — runs at 3am during storms where conditioning damage compounds. § Sourced
4. SAB-credentialed clinical claim — defensible protocol grounding Furbo and supplement vendors legally cannot match. ◌ Assumption

Product & BOM

Pawvlov M (hero SKU, 20–60 lb dogs) electronics BOM @ 10K volume totals $26.05/unit; assembly + test + inbound logistics adds $11.45 @ 10K → total landed COGS $37.50. Gross margin: S ($199) 81%, M ($249) 85%, L ($299) 86%. Tooling amortization ($105K / 3 mold sets over 10K units) adds ~$10.50/unit in year 1 and burns off after the first production run. △ Modeled

Renders

Three-size SKU strategy

Financial Model

Monthly cashflow exemplars

Risk Register

Engineering

Architecture

Embedded ESP32-S3 SoC running an onboard 4-class pose classifier (calm / alert / anxious / absent) at 5 fps plus an onboard MEMS-mic audio classifier (storm vs. non-storm) with ~80 Hz low-frequency trigger sensitivity. Optional cloud inference offload for edge cases; companion-app dashboards persist session history. The treat-delivery auger driver is locked behind the pose-classifier calm-window — never fires during a detected fear state. △ Modeled

Integrations

• Apple HealthKit-style pet-wellness export (read-only, optional)
• Veterinary behaviorist portal (SAB read-only access to session history for concierge MVP supervision)
• Standard 2.4 GHz WiFi (no Zigbee, no Matter at v1)

Data model

Core entities: Device, Dog (profile, weight, breed, anxiety triggers), Session (trigger event, classifier confidence timeline, treats delivered, calm-window count), Treat (size, jam events), Owner (subscription state, app prefs). Sessions are append-only and replayable for SAB supervision.

Security posture

Camera + audio inference happens on-device; video frames never leave the device (only classifier outputs do). Cloud inference path encrypts in transit and at rest; tenant isolation by owner_id row-level scoping. Consumer training aid, not medical device — no HIPAA/GDPR-health-record exposure. CARB + FCC Part 15B/15C pre-launch; UL listing year 2.

Scalability

Device fleet scales horizontally — each unit's inference burden lives on-device, so cloud cost grows only with subscription + session-history storage. Bottleneck above 100K units is the SAB review queue for concierge-supervised cohorts; mitigated by automated session-summary digests so DACVBs review exceptions, not every session.

Path to Product

Four phases, 18 months, $215,000. We don't commit mold-tooling budget until the classifier works in real homes. Every phase has a go signal and a rollback trigger. See /path-to-product for full detail.