How gantechproject/saie works
SAIE monitors SAP sources, turns changes into evidence-backed findings and ranked automation opportunities, and presents them in a dashboard and scheduled executive reports. The documented workflow spans discovery, crawling, analysis, scoring, reporting, delivery, governance, and retention. The tree and excerpts support the dashboard, API, persistence, LLM, scheduling, and delivery components; the detailed implementations of several agent stages were not sampled, so their internal wiring is shown only where documented.
Main entry points: schema.prisma Application database · agent-scheduler.ts Content analysis · App.tsx Dashboard · custom-routes.ts API routes
Components 19
Dashboard
- Dashboard
src/App.tsx - Login
src/components/LoginPage.tsx - Architecture viewer
src/App.tsx
API and agents
- HTTP server
server-node.ts - API routes
custom-routes.ts - Agent scheduler
agent-scheduler.ts
Analysis and reporting
- Source crawling
agent-scheduler.ts - Content analysis
agent-scheduler.ts - LLM client
llm-client.ts - Opportunity scoring
custom-routes.ts - Report generation
agent-scheduler.ts - Report delivery
report-delivery.ts
Data and operations
- Application database
prisma/schema.prisma - Database client
src/lib/db.ts - Data retention
data-retention.ts - Governance and audit
custom-routes.ts
Other components
- Decision maker
- SAP sources
- Delivery channels
Connections 21
- Decision maker → Login: signs in
- Login → HTTP server: submits credentials
- Decision maker → Dashboard: uses
- Dashboard → HTTP server: requests data
- HTTP server → API routes: dispatches requests
- API routes → Database client: queries and updates
- Database client → Application database: reads and writes
- Agent scheduler → Source crawling: schedules
- Source crawling → SAP sources: fetches pages
- Source crawling → Application database: stores crawl data
- Content analysis → Application database: reads crawl data
- Content analysis → LLM client: enriches findings
- Content analysis → Application database: stores findings
- Opportunity scoring → Application database: reads and records scores
- Agent scheduler → Report generation: schedules reports
- Report generation → Application database: reads report data
- Report generation → Report delivery: provides report
- Report delivery → Delivery channels: sends or exports
- Data retention → Application database: archives and purges
- API routes → Governance and audit: serves governance
- Dashboard → Architecture viewer: displays diagrams