Analyses completed
12,480
+384 in the last 24h
Face analysis control
Mirror the Expo flow on the web: enroll members, watch jobs move from pending to completed, and publish AI insights with the same HDKit-backed context the mobile app already trusts.
Capture uptime
Expo + web uploader healthQueue depth
7 analyses processing nowInsight SLA
< 90s avg AI handoffAnalyses completed
12,480
+384 in the last 24h
Avg processing time
68s
-14% vs last week
Uploads auto-retried
312
0 failed today
Guides monitoring
48
Ops + product pods
Platform
Capture on Expo, store through Nest, analyze via Python workers, and push insights back to HDKit-powered rituals without leaving this dashboard.
Camera overlay, pose checks, and contextual forms run inline with Expo + Next so every portrait ships with the metadata the backend expects.
Edge uploader compresses, encrypts, and streams to S3 via `/files/upload`, then mirrors presigned keys back to the reading record.
The Python face-analysis-service pulls from presigned URLs, runs MediaPipe landmarks, and reports status back through Nest + Prisma.
OpenAI summarization layers translate raw facial markers into coaching copy that slides directly into Readings, bonds, and rituals.
Status-aware UI mirrors Expo tabs so ops can trigger refreshes, pause jobs, or nudge guides before a member even notices delays.
Analyses sync back into HDKit-powered relationships. When a chart regenerates the right composite or ritual automatically follows.
Pipeline
Built from the FACE_ANALYSIS flow: Readings CTA, camera session, confirmation, asynchronous processing, and delivery back to every surface.
Step 1
Members tap Get Reading, see historical analyses, and get status-aware skeletons while new jobs spin up.
Step 2
Guided camera overlay ensures framing, auto-switches lenses, and captures the metadata the API requires.
Step 3
Users approve the shot, the client uploads via `/files/upload`, and a presigned key pairs with the person record.
Step 4
Nest marks the job PENDING → PROCESSING while the Python worker analyzes features and calls OpenAI.
Step 5
Once COMPLETED, Readings, bonds, and rituals receive the AI copy plus structured metrics for downstream automations.
Operations
These cards mirror the actual Prisma records and statuses the Expo clients poll so every pod sees the same truth.
Expo capture • Verified
Completed run with AI notes ready for publishing to Readings.
Web uploader • Pose pass
Face-analysis-service streaming data; OpenAI draft in ~45s.
Expo capture • Needs consent
Awaiting confirmation + S3 upload retry. Auto-pause engaged.
Desktop import
Pose failed twice. Worker escalated for review before retrying.
Solar plexus cues balanced; recommend grounding breathwork before daily ritual push.
Composite recalculating with latest markers. HDKit sync scheduled in 2 min.
Emotional wave flagged. Ops pinged to confirm consent + rerun.
Activity
Prisma triggers, hdkit recomputes, and the Python worker stream show up here so ops can triage without digging into logs.
New Expo capture synced (3.4 MB, checksum verified).
MediaPipe landmarks complete for Ezra · queued OpenAI step.
Draft ready for Maya’s ritual; awaiting guide confirmation.
Voices
Product, operations, and coaching leaders use this panel daily to mirror what the Expo app and backend services are doing.
“Face analysis finally shares a surface with our HD flows. I can check every queue without opening a terminal.”
“Status-aware widgets made ops calm again. If a job hangs I can rerun it directly from the dashboard.”
“Developers, guides, and support all point to the same Next.js app. Shipping rituals is half the time now.”
Compliance
Surface the same disclosures we run in Expo onboarding, legal docs, and the privacy center so every face-analysis job stays auditable.
Members approve the storage + processing of facial imagery inside the mobile confirmation screen and this dashboard mirrors that state.
Revoking analysis purges presigned links, invalidates OpenAI caches, and clears hdkit composites tied to that face ID.
We package US/EU biometric agreements, Apple ATT copy, and Google data disclosures directly into the onboarding modals.
Everything already lives in this monorepo—NestJS API, Prisma models, hdkit helpers, and the Python worker. This Next.js layer makes it operable for every team. Reach us anytime at info@auntoracle.com.
Join the waitlist
Share your best contact email and we will reach out with onboarding instructions, migration guides, and the next Cosmic Aunty release window.