Technical founder and behavioral systems architect. 18+ years in production software engineering, plus a psychology degree, building structured behavioral modeling infrastructure. Founder of Echos of Mind, a signal-based behavioral intelligence platform.
Echos of Mind is a deterministic behavioral signal engine that models longitudinal emotional patterns without relying on engagement manipulation to do it. Most mental health apps optimize for retention. This one optimizes for signal precision instead, which means it's built to need less from the user over time, not more.
The engine runs rolling window analysis across 24-hour, 7-day, and 30-day spans. Core detectors already implemented cover frequency spikes, volatility shifts, clustering, and intensity deviation, each with signal cooldown and deduplication enforced so nothing gets flagged twice. Execution runs on a cron-based, idempotent layer, and the whole thing runs on under $100/month in infrastructure.
Early-stage target: 1,000 real active users within 18 months, where active means three or more meaningful entries a month sustained for three months or longer.
Behavioral modeling has to be deterministic before it can be adaptive. The backend is modular NestJS with a Prisma schema-first data layer, indexed MySQL for window queries, a detector abstraction interface, and per-user baseline tracking. Access control is RBAC with granular permissions, token lifecycle is managed explicitly, and dev/prod parity runs through Docker. AI interpretation sits layered above the signal engine. It isn't embedded inside it, on purpose, so the core detection logic stays auditable and doesn't depend on a model's black box.
The system moves through three phases. It starts as structured journaling, where pattern detection runs on rolling deviations. From there it moves into adaptive calibration, refining per-user baselines and tuning sensitivity as it learns someone's actual range. The long-term shape is a behavioral operating system: longitudinal, multi-domain behavioral abstraction, where the compounding asset is how a person's baseline evolves over time, not how many entries they've logged.
Signal density matters more than streaks. Precision matters more than frequency. Restraint matters more than engagement pressure, and longitudinal modeling matters more than reactive feedback loops. A working version of this system should send fewer insights as it gets more accurate, not more.
Revenue has to track with depth of behavioral clarity, not attention capture. That points toward a premium adaptive calibration layer, longitudinal behavioral reports, and eventually institutional behavioral infrastructure. No ads, no retention engineering, not now and not later.
Clarity compounds when behavior gets structured over time. Echos of Mind starts as a journaling platform and evolves into adaptive behavioral infrastructure, deterministic at the core, adaptive at the edges.