Ingests an arXiv PDF URL and generates a VC-style investment report.
(Summary ➜ Market ➜ Competitors ➜ Startup Concept ➜ Investment Thesis)
- Framework: Next.js (App Router)
- Language: TypeScript
- Styling: Tailwind CSS
- UI Components: Shadcn UI
- Animations: Framer Motion
- API Runtime: Vercel Edge Functions
- AI: Google Gemini 2.5 Pro (via OpenAI SDK compatibility layer)
- Market Research: Firecrawl (
/deep-researchendpoint) - Competitor Analysis: Apify (Crunchbase Actor:
curious_coder/crunchbase-scraper) - PDF Ingestion: arXiv API +
pdf-parse+ guMCP
-
Clone the repository:
git clone <repository-url> cd paperchase
-
Install dependencies:
npm install
-
Set up environment variables: Create a
.env.localfile in the root directory by copying the example below. Obtain the necessary API keys and tokens:# .env.local # Google AI Studio API Key for Gemini access (https://aistudio.google.com/app/apikey) GEMINI_API_KEY= # Firecrawl API Key (https://firecrawl.dev/) FIRECRAWL_API_KEY= # Apify API Token (https://console.apify.com/account/integrations) APIFY_TOKEN=
-
Run the development server:
npm run dev-
Open http://localhost:3000 in your browser.
-
Paste an arXiv URL (e.g.,
https://arxiv.org/abs/2303.10130orhttps://arxiv.org/pdf/2401.08406.pdf) and click "Chase".
- Method: POST
- Body:
{ "arxivUrl": "<string>" } - Output: JSON representing the
InvestmentReport(seesrc/lib/types.ts).
- Implement more robust parsing for Firecrawl markdown output.
- Refine Apify Crunchbase Actor input/output mapping for better competitor data.
- Consider adding a Gemini call to analyze the competitor list from Apify.
- Add more sophisticated loading states (e.g., progress indicators for different steps).
- Improve error handling and user feedback.
- Add unit/integration tests.
- Implement diagram generation using VizCom or similar (original idea).
- Add authentication.
- Deploy to Vercel.