Reference: Financial Deep Research Agent
referenceresearch-deep
framework
Agent Framework
browser
Browser Automation
memory
Memory
retrieval
Retrieval / Vector Store
eval
Eval
observability
Observability / Tracing
sandbox
Sandbox / Code Execution
deployment
Deployment Target
mcp_servers
MCP Servers
Architectural notes
Reference architecture for a financial-research agent that produces earnings-call summaries, fundamentals comparisons, and thesis memos with traceable citations to filings. Why these picks: - Opus for the final synthesis where reasoning quality dominates. - Gemini for the "consume entire 10-K" pass. - EDGAR MCP for primary-source filings (no scraping). - Citation grounding on every claim — financial compliance requires it. - Langfuse self-hosted because thesis content is competitively sensitive. Common gotchas: - Hallucinated numbers are existential in finance; require citation + numerical-extraction confirmation. - Cost runaway is the operational risk; budget caps per research run.
Embed this stack
Show off your stack on your engineering blog or company page.
<iframe src="http://testserver/embed/stacks/research-deep-financial" width="100%" height="380" frameborder="0" style="border: 1px solid #334155; border-radius: 8px;"></iframe>