Mama Hala — AI-Native Counseling Platform
8 integrated AI systems plus a 12-phase self-driving booking subsystem powering a bilingual mental health practice — distributed slot lock, smart hold, auto-releasing unpaid slots, and a 2-surface admin (iPhone PWA + web) from one KV source of truth. Cut ~70% of admin load, freeing ~$150K/yr of founder time and replacing a $5-15K/yr SaaS stack.
ⓘProject description and metrics shown here are representative — drawn from a real engagement, with the client name anonymized where confidentiality applies.
It's 9:47pm and the practice owner is still at her laptop. Bookings lived in WhatsApp, invoices in Excel, reminders went out by hand, and unpaid sessions quietly turned into no-shows. Sensitive intake notes sat in an email inbox they never should have touched. Five disconnected tools, no single source of truth — and a practice that couldn't grow past one person's calendar and energy.
EiGENRA replaced that five-tool stack with one owned, AI-native engine that runs the practice itself — built to healthcare-grade PHIPA/PIPEDA security. Two layers fit together: eight specialized AI systems for the practice, and a self-driving booking subsystem that carries a client from intake to paid, reminded, and followed-up with no manual touch. Explore both in detail below.
A manual practice became a self-running operational engine. It replaced a modeled $5K–$15K/yr SaaS stack (Calendly, QuickBooks, Mailchimp, Twilio, a CMS) and cut admin workload roughly 70% — about 20 hours a week back to Dr. Hala. Eleven autonomous jobs run reminders, payment dunning, calendar cleanup, and retainer billing 24/7. Live at mamahala.ca.
This is one of the 8 AI systems, running right now. Describe what you're navigating — the AI matches you to a service from Dr. Hala's actual catalog.
This isn't a website. It's the operating system the practice runs on — one owned engine that swallows an entire stack of subscriptions:
70% of admin time reclaimed converts to new session capacity. Drag the sliders to model your own practice.
Every 'Book Now' click moves through 12 choreographed phases. Tap any step on the rail to see the timing, source files, and APIs behind it.
Bilingual landing + 4-question quiz routes the visitor to the right service and captures booker/attendee intent in Q1 (self / minor / couple).
While the owner sleeps, the engine runs the back office itself — eleven jobs, one rhythm. The conductor (booking-reminders) runs every 15 minutes and handles eight concerns in a single tick. Hover any spoke to read what it does.
The conductor — 8 branches in one tick
- 24h reminder (AI warmth + series prefix)
- 1h reminder + Meet link
- Pre-session payment nudges (configurable)
- Auto-complete at T+30min
- Auto-cancel at session-time + Stripe-grace
- Follow-up email at T+2h (audience-aware)
- Held-decision push for past-end confirmed-unpaid
- GCal event retry for missing calendarEventId
Each one chosen because it does its job quietly. The system is the choreography between them. Hover — or tap — any card to flip and see the detail and the in-code file path.
Each touchpoint is constrained — no diagnosis, no payment talk, no clinical homework. Claude writes one sentence of warmth, summarizes, matches — never replaces Dr. Hala.
Free-text concern → 3 ranked services in <3 seconds
AI-summarized intake — themes, tone, attendee facts
One affirming sentence in the 24h reminder, gender-neutral AR
One reflection sentence post-session, audience-aware
Admin records audio → AI transcribes + matches customer + drafts
120-word read-out on today's metrics for the admin home
Built to the hardest trust bar there is — mental-health data under PHIPA/PIPEDA. What holds up here holds up for your customers, your payments, your records.
Our principle: flagged, reversible, shipped in slices. Every capability launches behind a flag, gets smoke-tested live, and can roll back instantly — no big-bang launches, no surprises.