Enginuity Production Roadmap

Architecture & Deployment Sequence for "Future Max"

Phase 1: The "Dumb" Plumbing (Months 1-2)

Rule #1: Do not build AI yet. Build the data ingestion engine. If you can't reliably get the ERP data, the AI is useless.

  • Infrastructure: AWS S3 (Secure Storage) + AWS Lambda.
  • Action: Set up a secure SFTP folder. Tell your first pilot customer to schedule their legacy ERP (JobBOSS/E2) to drop the 4 standard CSVs here at 2 AM.
  • Processing: Write a simple Python script (Pandas) to clean the CSVs, standardize the headers, and calculate the CFO metrics (Margin Variance, CCC) using hardcoded math.
  • Frontend: Don't build a custom React app yet. Plug a BI tool like Metabase or Preset into your clean database and build the V2 CFO Dashboard there.

Phase 2: The Agentic Injection (Months 3-4)

Now that data flows reliably every 24 hours, you introduce the AI to generate insights, not just charts.

  • LLM Infrastructure: Azure OpenAI (Strictly for ITAR compliance and zero-training guarantees).
  • The Engine: Use a framework like LangChain or LlamaIndex. Write the "Trinity Graph" logic in Python.
  • Action: At 2:30 AM (after the CSV is processed), your Python script sends the statistical anomalies (e.g., "Weld Bay margin dropped 12%") to the LLM. The LLM returns a structured JSON string with the root cause and generative recommendation.
  • Output: The AI's insights are written back to your database and displayed on the dashboard next to the charts.

Phase 3: Multi-Tenant SaaS (Months 5-8)

Scaling from 1 pilot to 50 customers without your operations breaking.

  • Database: PostgreSQL with Row-Level Security (RLS) or entirely separate schemas per customer (Single-Tenant Data Vaults).
  • Frontend: Now you build the custom React/Next.js dashboard (the actual code we mocked up).
  • Onboarding Layer: Build the "Semantic Mapping" UI. When a new customer uploads their first CSV, a drag-and-drop tool lets your onboarding team map their columns to Enginuity's master schema.
  • Market Agent: Connect external APIs (commodity prices, vertical indices) to cross-reference against customer data.