Chinese market research is fragmented across filings, reports, and bilingual sources. Analysts spend substantial effort extracting text, reconciling evidence, and turning it into a consistent memo.
SilkRoad Agents coordinates three focused roles:
- Scout gathers and extracts source material, including Chinese PDFs through PaddleOCR.
- Analyst structures market, company, and risk evidence with ERNIE 4.5.
- Synthesizer turns the collected findings into a reviewable investment memo.
CAMEL-AI manages role-based collaboration. A FastAPI backend streams progress with server-sent events, while the Next.js interface keeps the source-to-memo workflow visible. The implementation also handles asynchronous HTTP requests, PaddleOCR API differences, and model-response normalization so failures remain inspectable.
The output is a structured research aid, not a guarantee of financial accuracy or investment performance. Analysts remain responsible for validating sources and conclusions before using the memo in a decision.
Built With
- camel-ai
- docker
- ernei-4.5
- fastapi
- javascript
- nextjs
- paddleocr
- python
Log in or sign up for Devpost to join the conversation.