|
|
|
|
AlphaDesk is a modern, modular financial analytics dashboard that bridges the gap between raw market data and actionable intelligence. Built with Next.js App Router and TypeScript, it aggregates real-time data from multiple providers and uses Generative AI (Gemini) to synthesize complex financial metrics into natural language investment memos.
- Real-Time Market Data: Aggregates live pricing, analyst ratings, and insider transactions from Yahoo Finance with sub-second latency.
- AI Analyst Agent: A custom RAG (Retrieval-Augmented Generation) pipeline that ingests 5+ distinct data streams to generate deep-dive equity research reports.
- Bento Grid Architecture: A modular, drag-and-drop interface built with
dnd-kitthat supports fully persistent layout personalization. - High-Performance Caching: Implements a tiered caching strategy (Upstash Redis) to mask expensive AI operations.
- Framework: Next.js 15 (App Router, Server Actions)
- Language: TypeScript
- Database: PostgreSQL (Prisma ORM)
- Caching: Upstash Redis (Serverless)
- AI Engine: Google Gemini 2.5 Flash via Vercel AI SDK
- Styling: Tailwind CSS, Shadcn/UI
- State Management: React Server Components (RSC) + Suspense Streaming
One of the core challenges of AlphaDesk was balancing the high latency of Generative AI (RAG pipelines) with the need for a snappy user experience.
The Problem: Generating a deep dive requires fetching multiple data modules like Insiders, Financials, News, Ratings etc. and running a complex LLM prompt.
- Average Cold Load Time: ~8.8s
The Solution: Implemented a Stale-While-Revalidate caching strategy using Upstash Redis to serve cached insights instantly while updating in the background.
- Average Warm Load Time: <150ms
- Latency Reduction: ~98%