Show HN: 6 Months Building and 2 Refining – I Shipped My AI Accountability Coach (justgrinds.vercel.app)

An indie developer shipped an "AI Accountability Coach" that blends conversational coaching, habit tracking, and affective logging to help users build consistent routines. The app asks structured diagnostic questions, lets you add habits and track streaks, logs emotions (fear, motivation, regret, etc.), and computes an automated "consistency" score. Every response is tailored to your current streak and emotional state—offering personalized prompts, gentle re-entry messages after a slip, and adaptive nudges rather than generic pep-talks. Core features are available without a paywall, with a Pro tier for advanced usage; early users report longer streaks and lower stress by replacing multiple apps with one context-aware chat. For the AI/ML community this is a compact example of applied adaptive conversational systems and affective computing: it operationalizes longitudinal user state (streaks, mood trends) into dynamic message generation and metric-driven feedback loops. Key technical implications include real-time user-state inference, personalized prompt generation, and consistency-scoring as an optimization objective—areas ripe for experimentation with model conditioning, reward signal design, and privacy-preserving telemetry. It also surfaces ethical and evaluation questions (over‑nudging, dependence, sensitive emotion data) that researchers should address when deploying behavior-change AIs in the wild.