Path to Product

From POC to shipped

Four phases from validation to shipped product, each gated by an explicit go/no-go signal. The classifier-validation study is the gate that controls everything downstream — nothing scales until it clears.

Phase 1 — Active

Validate the classifier (the risk that gates existence)

Collect 1,000+ labeled dog-behavior frames across breeds, lighting conditions, and angles including docked-tail and cropped-ear edge cases. Validate ESP32-S3 on-device inference at greater than 95% calm/anxious accuracy with under 2% false-positive fear-reward on held-out data. Run a concierge behaviorist-supervised beta in 5 homes (log-only, no dispense). Build the storm-belt waitlist.

Timeline
75 days
Cost
$26,000
Go signal
>95% calm/anxious accuracy + <2% false-positive fear-reward on held-out data AND 500 storm-belt waitlist signups

Phase 2

Engineering prototype — single M SKU, on-device architecture

Schematic + PCB rev 1–3. FreeRTOS firmware with the firmware-locked calm gate. On-device-only inference (no cloud pose data — BIPA mitigation). Validate the <40 dB auger across treat-size variance in end-of-line testing. Mobile app v0. Commit the single-cavity $35K injection-mold tool. FCC pre-scan at design freeze.

Timeline
120 days (8–12 wk mold lead time)
Cost
$72,000
Go signal
5 working prototypes pass <40 dB dispense + jam-recovery + fail-closed gate; mold first-shots dimensionally in spec

Phase 3

Certify, comply, pilot

FCC Part 15B/15C + UL 62368-1 + CARB. Privacy and legal build-out: two-party-consent disclosure for CA/FL/IL/MA/PA/WA, BIPA on-device architecture opinion, auto-renewal disclosure. First 1,000-unit component buy and production run. Storm-belt pilot launch to the waitlist.

Timeline
90 days
Cost
$62,000
Go signal
FCC + UL granted; privacy legal opinion clean; 250+ paid pre-orders convert from the waitlist

Phase 4

Scale + prove the economics

Add S/L SKUs via family-mold inserts. Scale storm-belt DTC with a real CAC budget. Grow subscription attach. Instrument LTV/CAC. Target Series A metrics.

Timeline
6 months
Cost
$60,000
Go signal
5,000 cumulative units, >40% subscription attach, CAC <$150, line-of-sight to $50K MRR

Capital Required

Total raise: $245,000. The raise is right-sized to include CAC (the line the $215K founder-plan estimate omitted), classifier-data collection, engineering prototype through certifications, and a veterinary-behaviorist SAB retainer. The single-SKU decision (Medium only at launch) redirects ~$70K that would have gone to a three-cavity mold into the things that determine whether this product should exist: the classifier study, certifications, and actual customer acquisition.

See the full use-of-funds breakdown →

Current status

As of 2026-07-10

Completed

  • PE evaluation complete (routing FUND, pe_score 7.1)
  • Product design + electrical spec complete (single M-SKU, on-device architecture, <40 dB auger, floor-reachable tray)
  • Competitive research + honesty flags complete
  • POC demo site built and live at pawvlov3.nltlabs.ai

Pending

  • Phase 1 classifier-validation study (the gate to everything downstream)
  • Seed round ($245K, right-sized to include CAC + data collection + SAB)
  • DACVB behaviorist SAB recruitment (3 members targeted)
  • Storm-belt waitlist build (target: 500 before Phase 2 commit)