Project Delta: Architecture & Context Overview¶
❖ Project Summary¶
Codename: Project Delta Type: Off-Salesforce Spinoff Target Launch Region: India Target Users: Indian film producers with greenlighting authority
Purpose: To deliver a lightweight, scalable, and AI-enabled film production finance tool outside the Salesforce ecosystem. Project Delta provides a simplified set of features originally built for UK-based tax credit compliance, now re-engineered to suit a global, price-sensitive SaaS market. The project's phased purpose is:
- Phase 1: Enable financial performance tracking and budget adherence during film production.
- Phase 2: Enable assessment of film project viability and profitability.
❖ Strategic Rationale¶
The legacy Salesforce-based platform served UK film productions focused on tax credit workflows. However, expansion into new markets (e.g., India) requires an off-platform solution due to:
- Salesforce Licensing Limitations: Community licensing is expensive and inflexible for SaaS.
- AI Integration Challenges: Limited ability to integrate AI frameworks and LLMs in Salesforce.
- UK-Specific Architecture: Existing system heavily customised for UK tax credit compliance.
- Limited Framework Flexibility: Salesforce ecosystem restricts integration with modern tools.
❖ Target Audience¶
Indian film producers with greenlighting authority.
❖ Technical Architecture¶
⚙️ Tech Stack¶
App¶
- Framework: Next.js
- Hosting: Vercel (Hobby / Pro)
- Design: Mobile-first, responsive UI
Database¶
- Platform: Supabase
- Database: PostgreSQL
- Auth: Supabase Auth (email/magic link), Twilio/WhatsApp
- Edge Functions: Supabase Edge Functions (for WhatsApp and lightweight APIs)
AI Hub¶
- Languages: TypeScript / Python
- Orchestration: Genkit (running in AWS ECS Fargate)
- Fast Jobs: API Gateway → ECS
- Long Jobs: AWS SQS → Fargate
- Secrets: AWS SSM Parameter Store
Monitoring¶
- Logging: Grafana Cloud (Free Tier – 50GB/month)
Tooling¶
- Project Management: Linear
- Source Control: GitHub
- AI-first Dev: via Cline + MCP
Messaging¶
- Bot Interface: WhatsApp bot via Supabase Edge Function
❖ Core Features by Phase¶
Phase 1: Foundation (Live)¶
- Import budget in PDF/CSV (AI structured parsing).
- Import expenses/invoices via WhatsApp, email or import directly into app dashboard.
- Simplified user onboarding (WhatsApp-based for producers and team members).
- OCR + LLM pipeline for invoice ingestion.
- Auto-link expenses to budget line items.
- Transaction review interface (human-in-the-loop for OCR and linking review).
- Table-based reporting: grouping, filtering, sorting.
- Dashboard: multiple projects, drill-down enabled.
- Visualisation layer: charts with drill-to-table interaction.
Phase 2: Production Levers (In Development)¶
- Script-to-budget automation (rapid conversion with dynamic financial impacts).
- Scene creation, production item suggestion (including AI-generated imagery for production requirements), and schedule generation from budgets.
- Real-time budget updates from production schedule/scene edits.
- Structured reporting to answer:
- “Will I make money?” (addressing financial viability)
- “Am I on track?” (addressing project profitability)
Phase 3: Presentation (Planned)¶
- AI-generated storyboards from scenes.
- AI-generated pitch packs from script.
❖ Web App – Key System Requirements¶
- WhatsApp Authentication: Simplified onboarding and user verification.
- Multicurrency Support:
- Fixed-rate exchange mapping per project.
- Multi-currency display UI (user-chosen overlay alongside native value).
- Subscription & Payments:
- Stripe integration for SaaS + pay-per-use.
- Credit balance system.
- Permissions & Roles:
- User-level permission control within projects.
- Multilingual UI:
- Internationalisation architecture from day one.
- Dark/Light Mode Toggle: User-controlled interface theme options.
❖ Summary for LLM Context¶
Project Delta is a lean, AI-integrated web application for Indian film producers to manage production finances, link real-world expenses to planned budgets, and dynamically adjust budgets based on live production variables. It is a breakaway from a UK Salesforce-based system, now rebuilt on a modern, open stack (Supabase, Next.js, Genkit, AWS). The new app must be fully usable via mobile/WhatsApp, support local currencies, and integrate tightly with AI/LLM workflows for data extraction, budget modelling, and visual reporting.
All code, documentation, and AI behaviour must assume zero Salesforce involvement, and full architectural independence. Use Cline for bot-enhanced dev cycles and Linear for project management.
❖ Related Repositories & Documentation¶
sanaApp/: original business logic engine (deprecated, Salesforce)aihub/: Node.js + Genkit AI orchestration layer (pertains to Project Delta's AI hub)dashboard-demo/: legacy UI/UX patterns from Experience Cloud (reference for understanding historical UI, not for direct replication)oldCompanyAndServiceOverview.md: background on Farm Cove’s evolution (provides historical business context)oldCodebaseSummary.md: details of deprecated system components (useful for understanding what not to replicate from the legacy system)oldTechStack.md: legacy Salesforce + AI architecture and analysis tools (provides context on previous technical decisions and limitations)
❖ Project Vision¶
Project Delta aims to rapidly become the preferred financial management and viability assessment tool for film producers in India, leveraging modern AI, scalable technology, and user-centric design to significantly outperform traditional Salesforce-based solutions.