StrategyRoom.ai

AI-powered investment research workspace that blends live web intelligence, structured financial analysis, portfolio watchlisting, and sector-wide scanning.

Quick Navigation

• Experience Snapshot
• What and Why
• Interactive Architecture Map
• Feature Tour
• Tech Stack Ribbons
• Run Locally
• Environment and Secrets
• Data Model and Security
• Scripts
• Testing
• Deployment Notes

Experience Snapshot

StrategyRoom.ai behaves like a compact research command center:

1. Search any company or ticker.
2. Watch a live two-agent timeline as research and analysis run.
3. Inspect a structured report with verdict, confidence, fundamentals, sentiment, risks, reasoning chain, and citations.
4. Save to watchlist, set alert rules, compare multiple names, and export reports.
5. Re-research runs on schedule and notifies you when key signals change.

What and Why

What it does

It converts a plain-language query into an explainable investment report with consistent structure:

• Verdict: Buy, Hold, Sell
• Confidence score plus confidence breakdown factors
• Executive summary
• Fundamentals and data visualizations
• Sentiment analysis and analyst consensus signal
• Risk factors
• Source links and AI reasoning chain

Why it exists

Equity research is often fragmented across news sites, reports, and dashboards. StrategyRoom.ai is designed to close that gap by making research:

• Faster: automated source collection and synthesis
• More explainable: reasoning chain and confidence decomposition
• More actionable: watchlist persistence, alerts, and comparison views
• More continuous: scheduled re-research and notifications

Interactive Architecture Map

flowchart LR
    A[User Query] --> B[Frontend Dashboard]
    B --> C[Research API Orchestrator]
    C --> D[Edge Function: researcher]
    D --> E[Firecrawl Search]
    C --> F[Edge Function: analyst]
    F --> G[Gemini 2.5 Flash]
    F --> H[Structured Report JSON]
    H --> I[Report UI and Charts]
    I --> J[Watchlist and Alerts]
    J --> K[Supabase Postgres]
    K --> L[Cron Re-Research]
    L --> D
    L --> F
    L --> M[Alert Notifications]

Loading

Feature Tour

Core flows

• Auth flow with Google OAuth and email/password
• Company research and structured report generation
• Sector scan across top companies with confidence ranking
• Watchlist cards with quick refresh and compare actions
• Alert settings for verdict change, sentiment shift, confidence drop
• Notification bell with unread tracking
• Export as PDF and PNG, plus quick summary copy

Primary app routes

• App shell and protected routing: src/App.tsx
• Main dashboard: src/pages/Index.tsx
• Comparison mode: src/pages/ComparisonView.tsx
• Sector analysis: src/pages/SectorView.tsx
• Login flow: src/pages/LoginPage.tsx

Key feature modules

• Report composition: src/components/ReportPanel.tsx
• Live activity timeline: src/components/AgentActivityFeed.tsx
• Watchlist interaction: src/components/WatchlistPanel.tsx
• Alert controls: src/components/AlertSettings.tsx
• Notifications panel: src/components/NotificationBell.tsx
• Export controls: src/components/ExportToolbar.tsx

Tech Stack Ribbons

Frontend ribbon

Backend and AI ribbon

Quality and delivery ribbon

Note on Python: this repository is TypeScript-first across frontend and edge functions, with no Python runtime used in the current codebase.

Run Locally

Prerequisites

• Node.js 18+
• npm 9+
• A Supabase project (cloud or local)
• Firecrawl and Gemini API keys

Quick start

npm install
npm run dev

Default local app port is 8080.

Environment and Secrets

Frontend environment variables

Create a .env file in project root:

VITE_SUPABASE_URL=your_supabase_url
VITE_SUPABASE_PUBLISHABLE_KEY=your_supabase_anon_key

Supabase edge function secrets

• FIRECRAWL_API_KEY
• GOOGLE_API_KEY
• SUPABASE_URL
• SUPABASE_SERVICE_ROLE_KEY
• CRON_SECRET (recommended)

Data Model and Security

Tables

• research_history
• watchlist
• alert_settings
• alert_notifications

Security model

• Row-Level Security is enabled on user-owned tables.
• Policies ensure each user can only read and write their own records.
• Unique constraints prevent duplicate ticker entries per user in watchlist and alert settings.

Migrations

• supabase/migrations/20260220_research_history.sql
• supabase/migrations/20260222_watchlist.sql
• supabase/migrations/20260223_alerts.sql

Scripts

• npm run dev: start development server
• npm run build: production build
• npm run build:dev: development-mode build
• npm run preview: preview production output
• npm run lint: run ESLint
• npm run test: run tests once
• npm run test:watch: run tests in watch mode

Testing

Current tests validate:

• Financial parsing logic for chart metrics
• Glossary term completeness and lookup normalization

Test files:

• src/test/charts.test.ts
• src/test/glossary.test.ts

Deployment Notes

• SPA rewrite is configured in vercel.json.
• PWA support is enabled with vite-plugin-pwa.
• Current function config includes verify_jwt = false for several edge functions; production hardening should enforce stricter invocation controls and secret-gated cron access.

Final Product Summary

StrategyRoom.ai is an explainable AI research cockpit for equity decisions. It does not just generate one-off output; it supports the full lifecycle of research, monitoring, and signal-change awareness.