💡 Voice command practice mode for zero-tap field readiness onboarding

[github-actions[bot]]

Summary

An interactive sandbox where new beekeepers rehearse Broodly's voice commands with simulated hive scenarios before their first field session. Walk through a virtual 3-hive inspection: practice "next hive," "brood pattern looks good," "take photo," and hear the system's TTS responses. Reduces first-use anxiety, improves voice recognition accuracy through user-specific voice profiling, and builds confidence in the zero-tap workflow before gloves go on. Accessible from onboarding flow and settings for seasonal refresh.

Market Signal

Tunen, a voice-based agriculture field assistant, invests heavily in domain-specific language training to improve field recognition accuracy — validating that voice practice directly improves real-world performance. No beekeeping app offers a practice mode for voice interaction. HiveSense ships on-device Whisper-based voice features but provides no onboarding — users must figure out commands in the field under pressure. The broader voice assistant industry (Alexa Skills, Google Actions) has demonstrated that users who try voice commands in low-stakes settings use them 3-5x more frequently in real contexts.

User Signal

The PRD's primary user journey (Newbie Hannah) explicitly identifies anxiety as the starting emotional state: "anxious about doing harm by doing the wrong thing." The emotional arc is anxious → guided → confident. A practice mode directly accelerates this arc by letting Hannah try voice commands at her kitchen table without the pressure of live bees, smoke, and gloves. The PRD's time-to-first-recommendation target (median under 10 minutes from account creation) would benefit from a fast, engaging onboarding that doubles as voice system calibration. No existing idea in the backlog addresses voice onboarding or practice.

Technical Opportunity

The voice capture pipeline (Epic 11, Stories 11.2-11.3) and TTS system are already planned. A practice mode reuses the same STT/TTS infrastructure with simulated inspection state instead of live data. The Zustand inspection store can be initialized with mock hive data (3 hives with pre-set states) for practice sessions. Expo Router can host the practice flow as a standalone route (/practice) accessible from onboarding or settings. The practice mode runs entirely on-device — no backend infrastructure needed. Voice profiling data captured during practice (accent, pace, terminology preferences) improves STT accuracy for subsequent real inspections.

Assessment

Dimension	Score	Rationale
Feasibility	high	Reuses existing STT/TTS infra with mock data; no new backend; runs entirely on-device
Impact	med	Reduces onboarding friction for core differentiator; improves voice recognition accuracy; increases feature adoption
Urgency	med	Should ship with or soon after the voice inspection flow to maximize adoption of zero-tap mode

Adversarial Review

Strongest objection: Building a practice mode is engineering effort spent on a feature that only gets used once or twice per user. The ROI is questionable when that development time could go toward improving the actual inspection flow.

Rebuttal: The practice mode serves three purposes beyond one-time use: (1) Voice recognition calibration — the system learns the user's accent, speaking pace, and terminology preferences (some say "super," others say "honey box"), improving field accuracy for every subsequent inspection. (2) Feature discovery — users who practice use 3-5x more voice commands in the field, increasing engagement with Broodly's core differentiator. (3) Seasonal re-engagement — returning users who haven't inspected in months can refresh their command vocabulary before spring, reducing churn. Implementation cost is low because it reuses existing STT/TTS infrastructure with mock data rather than building new systems.

Suggested Next Step

Design 3 practice scenarios of increasing complexity: (1) Basic navigation — "next hive," "previous hive," "done" with system confirmation. (2) Observation logging — "brood pattern good," "queen spotted," "low stores" with context-appropriate TTS responses. (3) Full mini-session — "start inspection," navigate 3 hives, log 2-3 observations per hive, "end session." Define the voice command vocabulary that the practice mode teaches (target: 15-20 commands covering navigation, observations, photos, and session management).