AgentForger

Case Study

AI Research Agent for a Trading Firm

How AgentForger built personalized research agents that process announcements, filings, earnings calls, PDFs, and web sources for traders.

Prop trading firmUS / Hong Kong

The problem

A workflow with too much manual handoff

Traders needed to monitor announcements, IPO filings, earnings calls, PDFs, and web sources without relying on generic chat tools that missed context or hallucinated.

Research involved manual monitoring across fragmented sources.

Analysts repeatedly summarized the same source types for different traders.

Generic chat tools lacked trader-specific context and source discipline.

What we built

A controlled agent workflow

AgentForger built personalized research agents that process internal and external data, draft investment memos, cite sources, and adapt to each trader's research style.

  • RAG
  • web research
  • PDF processing
  • memo drafting
  • source citation

Results

  • ~10 hrs
    Saved per trader weekly
  • $100K
    Equivalent time saved firm-wide

The research workflow became faster and more personalized, with better source grounding and less repeated manual summarization.

FAQ

Questions about this workflow pattern

Can this type of agent cite sources?

Yes. Research agents should cite source material and separate retrieved evidence from model-generated synthesis.

Is this financial advice?

No. The workflow supports research and memo drafting. Final investment decisions remain with qualified humans.

Build around one workflow

Tell us what your team is still doing manually.

WhatsApp us