How is sensitive financial data handled?
The workflow is built with scoped access and privacy boundaries so confidential portfolio and transaction data stays controlled, with human review on advice.
Case Study
AI Investment Agent for a Hong Kong Family Office
How AgentForger built a portfolio tracker and AI investment agent that manages PDF transaction records, monitors position events, and advises against the family office's own strategy.
Client: Golden InvestmentFamily officeHong Kong
The problem
A workflow with too much manual handoff
A Hong Kong family office tracked a complex portfolio across PDF transaction records and structured products. Position events and underlying news were monitored manually and at risk of being missed.
Transaction records lived in PDFs that had to be read and reconciled by hand.
Structured-product events such as knock-out, strike, and settlement were tracked manually.
Investment views were not consistently checked against the family office's own strategy and performance.
What we built
A controlled agent workflow
AgentForger built an investment portfolio tracker and an AI investment agent that ingests PDF-based transaction records, monitors position events such as knock-out, strike, and settlement, follows underlying news, and provides investment advice grounded in the family office's actual strategy and portfolio performance.
- PDF processing
- portfolio tracking
- event monitoring
- news monitoring
- strategy-aware advice
Results
- PDF → portfolioAutomated record tracking
- Event alertsKO, strike, settlement, news
The family office gained a live view of its portfolio, automatic alerts on position events, and advice grounded in its own strategy instead of generic market commentary.
FAQ
Questions about this workflow pattern
Does the agent make investment decisions?
No. It tracks positions, monitors events and news, and drafts advice grounded in the client's strategy. Final investment decisions remain with the family office.
Build around one workflow