Agentic AI
AI Portfolio Assistant
A conversational investment portfolio agent that generates personalised rebalancing recommendations based on real-time market data and user risk profiles.
- Industry
- Wealth management
- Status
- Delivered
- Outcome
- 28% increase in user engagement compared with rule-based systems
Overview
A conversational agent that explains portfolio positioning and proposes rebalancing steps within the constraints of a stated risk profile.
The challenge
Static rule-based guidance gave the same output regardless of context, so users stopped engaging with it.
Architecture
- LangGraph agent graph separating data retrieval, analysis and drafting
- Alpaca API integration for market and position data
- GPT-4 explanation layer bounded by the retrieved portfolio state
- React conversational interface with proposal review before action
Capabilities delivered
- Conversational portfolio review
- Risk-profile aware rebalancing proposals
- Real-time market data grounding
- Explanations attached to each recommendation
Outcome
28% increase in user engagement compared with rule-based systems.
More work