- Implemented Citations component to display citation data. - Created WatchlistDrawer and ChatSidebar components for managing watchlists and AI chat functionality. - Integrated API calls for watchlist management and chat interactions. - Updated index.tsx to include new components in the main application layout. - Added API client in lib/api.ts for structured API interactions. - Developed Alerts, Dashboard, Portfolio, Routines, Screener, and Report pages with relevant data fetching and UI components. - Introduced styles in tokens.css for consistent theming across the application. - Configured TypeScript and Vite for project setup and development.
1.4 KiB
FlowSight — Demo Deck (offline, seeded)
Narrative: briefing → radar alert → report → interrogation. All numbers cite snapshots; stale chips render marked, never hidden.
Slide 1 — Problem
6M+ retail SID decide from price + rumor. Institutional data (broker flow, foreign flow, insider filings) is raw across 70 Sectors endpoints.
Slide 2 — Autopilot routines
Subscribe once; agents run on schedule; results land without opening the app. R1 Morning Briefing 07:30 · R2 Radar 30-min · R3 Reversal · R4 Insider Tape · R5 Earnings H-7/H-3/H-1 · R6 Dividend H-14/H-3 · R7 Weekend Review.
Slide 3 — Live: Morning Briefing
GET /api/briefing/today — top-5 accumulation, foreign table, 7d agenda,
mover of the day. ≤25 lines, every row cited.
Slide 4 — Live: Accumulation Radar
BBCA: 3 brokers net-buy Rp1.2T, volume 3.2× 20d avg → alert card with score,
top-3 brokers, volume multiple, /report/BBCA link.
Slide 5 — Live: One-Click Report
POST /api/report/BBCA — 7 sections with citations[] per section; export
PDF/HTML/MD/JSON. Thesis claims carry inline [endpoint @ date] markers.
Slide 6 — Verifiable AI + Accuracy Ledger
Every number cites endpoint + snapshot timestamp. Recommendations resolve at
+30d (ResolveDue daily cron); agent weights follow hit-rate, shown in
/portfolio accuracy table.
Slide 7 — Ask
Run BASE=http://localhost:8080 sh scripts/demo.sh offline from seed.