AI Portfolio Construction Digital Worker
Deploys a multi-agent AI workflow using the ReAct (Reasoning + Acting) pattern that automates fund research across 300,000+ instruments, applies ESG screening with SFDR compliance, performs mean-variance portfolio optimization, executes Monte Carlo risk analysis, optimizes UK tax wrappers, and validates MiFID II/PRIIPs regulatory requirements in minutes rather than hours..
Problem Statement
The challenge addressed
Solution Architecture
AI orchestration approach
AI Agents
Specialized autonomous agents working in coordination
Orchestrator Agent
Coordinating multiple specialized AI agents to work together cohesively on complex portfolio construction tasks requires centralized management of task decomposition, agent delegation, and result synthesis.
Core Logic
Uses Claude 3 Opus to decompose client requirements into discrete tasks, delegates work to specialist agents based on capabilities, monitors workflow progress across all phases (initialization, data gathering, analysis, optimization, validation, synthesis), and synthesizes final portfolio recommendations with executive summaries. Maintains working memory and coordinates inter-agent communication.
Market Intelligence Agent
Investment decisions require current understanding of market conditions, economic indicators, sector trends, and market regime (risk-on/risk-off) which constantly change and require significant research effort.
Core Logic
Analyzes real-time market indices (FTSE100, S&P500, MSCI World), economic indicators (GDP, CPI, unemployment, interest rates), sector momentum and rotation signals, and volatility regimes. Determines current market conditions to inform portfolio construction decisions with sector overweight/underweight recommendations.
Fund Research Agent
Manually searching through 300,000+ funds to find those matching specific client criteria (cost, performance, domicile, sector exposure) is extremely time-consuming and may miss optimal options.
Core Logic
Connects to fund database, applies multi-criteria filters (OCF limits, minimum fund size, domicile requirements, sector exclusions), ranks funds by Sharpe ratio and risk-adjusted returns, retrieves comprehensive fund data including holdings, and maintains shortlists of top-performing candidates for further analysis.
ESG Analyst Agent
Screening investments for environmental, social, and governance criteria requires expertise in multiple ESG frameworks, access to ESG ratings data, and understanding of SFDR Article 6/8/9 classifications.
Core Logic
Loads MSCI ESG ratings for shortlisted funds, applies sector exclusion screens (fossil fuels, tobacco, weapons, gambling), validates SFDR Article classification compliance, checks carbon intensity limits, and makes data-driven decisions on whether to apply strict (Article 8/9 only) or flexible screening based on client ESG preferences.
Risk Analyst Agent
Portfolio optimization requires sophisticated quantitative analysis including correlation matrices, mean-variance optimization, and stress testing which demands specialized financial engineering expertise.
Core Logic
Calculates pairwise correlation matrices for all shortlisted funds, runs Markowitz mean-variance optimization to find the efficient frontier, performs Monte Carlo simulations (10,000 scenarios) for probability distributions, applies historical stress tests (2008 crisis, COVID-2020, 2022 inflation), and generates three portfolio variants (conservative, balanced, aggressive) with risk metrics.
Tax Optimization Agent
UK investors can significantly reduce tax burden through optimal allocation across ISA, SIPP, and GIA wrappers, but calculating the optimal strategy requires knowledge of current allowances, contribution limits, and tax relief rules.
Core Logic
Retrieves current UK tax allowances (ISA £20,000, Pension £60,000 annual allowance, CGT £3,000), calculates optimal wrapper allocation based on client circumstances and time horizon, models tax relief benefits (40% higher-rate pension relief), develops capital gains tax crystallization strategies, and projects lifetime tax savings.
Compliance Agent
Investment recommendations must comply with MiFID II suitability requirements, PRIIPs KID availability, SFDR disclosures, and complete cost transparency, requiring detailed regulatory knowledge and verification.
Core Logic
Validates all three portfolio options against MiFID II suitability requirements (risk alignment, knowledge & experience, financial situation, investment objectives), verifies PRIIPs KID document availability for all funds, confirms complete cost disclosure across OCF, transaction costs, and performance fees, and generates comprehensive audit trails with cryptographic hashes.
Sentiment Analyst Agent
Market sentiment from news and social media can provide valuable signals for investment timing, but manually monitoring and analyzing thousands of sources is impractical.
Core Logic
Processes 500+ financial news articles from Reuters, Bloomberg, FT, and CNBC using NLP sentiment analysis. Analyzes social media sentiment from Twitter/X, Reddit, and StockTwits. Calculates aggregate sentiment scores, identifies key market-moving themes, and detects Fear & Greed index levels to inform portfolio timing recommendations.
Worker Overview
Technical specifications, architecture, and interface preview
System Overview
Technical documentation
Tech Stack
7 technologies
Architecture Diagram
System flow visualization