A production AI platform that turns a name and birth date into deeply personalized numerology reports across four products, generating multi-section PDF reports in 33 languages with Google Gemini and a custom RAG knowledge base.
Commercial product — source code is private.

Soul Aura is a commercial AI platform built on Pythagorean numerology, Jungian archetype psychology, and North-Node astrology. It generates personalized, multi-section PDF and Markdown reports from nothing more than a name and a date of birth. The platform spans four specialized products — Spiritual, Career, Relationships, and Wealth — each running as an independent Dockerized microservice on a shared network. Every report is produced by Google Gemini, grounded section-by-section in a hand-crafted Retrieval-Augmented Generation (RAG) knowledge base, validated by automated QA controls, and delivered by email through fully automated Shopify and n8n workflows.
Four specialized report products — Spiritual, Career, Relationships, and Wealth — sharing one infrastructure
Personalized multi-section PDF and Markdown reports generated from just a name and date of birth
Retrieval-Augmented Generation (RAG) grounds every section in a hand-crafted numerology knowledge base for accuracy
Google Gemini orchestration with section-by-section prompt building and motor self-checks
33-language localization with culturally adapted content
Automated QA validators (20-34 named controls) catch hallucinations and unsafe claims before delivery
Shopify to n8n to report pipeline turns a paid order into a delivered PDF automatically
Transactional report delivery via Resend
Tiered commercial offering (Basic, Premium, VIP) with per-tier section depth
Deterministic calculation engine (Pythagorean numerology plus Lunar Nodes) kept separate from AI generation
Dockerized microservices on a shared network for independent scaling
Hundreds of automated tests (e.g. 169/169 on the Relationships service) guarding every release
Soul Aura is a complete, production-grade numerology platform that combines deterministic numerological computation with grounded generative AI to produce premium, personalized reports at scale. Rather than letting a language model free-associate, the system first computes an exact numerological profile with a dedicated calculation engine, then uses Retrieval-Augmented Generation to ground every Gemini-written section in a curated knowledge base — keeping the output accurate, on-brand, and safe. The platform is structured as four parallel products that share one infrastructure: Spiritual (16-section transformation reports), Career (15-section guidance for students and young adults), Relationships (tiered couples reports), and Wealth (numerological wealth profiling with an AI-coach package). Each product is an independently deployable Flask service on a shared Docker network, reusing the same email backend, fonts, and cross-theme QA helpers. What makes Soul Aura genuinely commercial is everything around the model: a 33-language localization layer with cultural adaptation, automated QA validators with dozens of named controls that catch hallucinations and over-claims before delivery, hundreds of automated tests guarding content quality across releases, tiered pricing, and a hands-off Shopify to n8n to report pipeline that turns a paid order into a delivered PDF with no human in the loop.
Four independent Flask and Gunicorn services (Spiritual, Career, Relationships, Wealth) each run in their own Docker container on a shared n8n-network, reusing common fonts, the Resend email backend, and a shared QA helper library imported as a Python package. A single symlinked environment file is the one source of secrets across all services, so rotating a key updates every product at once.
Each report starts with a deterministic calculation engine that computes the full numerological profile. A prompt builder then extracts the relevant RAG chunks per section and assembles a Gemini prompt; the model writes the section; and a motor self-check plus a multi-control QA validator verify the output before the report is assembled. This calc-then-RAG-then-generate-then-validate flow keeps results accurate and on-brand.
Large curated RAG corpora — hundreds of chunks per product (e.g. a v22 relationship RAG of 601 chunks, a 14k-line wealth knowledge base) — encode Soul Aura's proprietary numerology content. A localization layer renders reports in 33 languages with cultural adaptation rather than raw machine translation.
A typography-aware PDF generator produces premium, multi-section reports with custom fonts, alongside Markdown export and shareable summary cards. Rendering is decoupled from generation so the same content can be delivered in multiple formats.
n8n workflows connect Shopify orders to the report APIs (including dedicated student and dershane flows), trigger generation, and deliver the finished report by email — a fully hands-off path from purchase to delivery.
Per-product test suites and named QA controls (prophecy, over-claim, and safety checks among them) gate every release. Versioned changelogs, compliance certificates, and deployment briefs keep the four services releasable in parallel without regressions.