Global wealth management firm deploys Squirro's generative AI employee agents to help 900 client advisors navigate regulatory compliance

Background

In wealth and asset management, regulatory complexity is high and increasing. Client advisors must navigate changing regulations across multiple jurisdictions while simultaneously managing client relationships and making investment decisions. New advisors face a steep learning curve; experienced advisors face compliance drift risk as regulations evolve. The volume and fragmentation of institutional knowledge — across SharePoints, policy manuals, compliance frameworks — creates a significant information bottleneck that generative AI is well-suited to address.

What Was Implemented

• Squirro Chat deployed as a generative AI employee agent across 900 client advisors
• RAG-based architecture connecting 15 SharePoint instances plus HR data, market data, FactSet, regulatory frameworks, policy manuals, training documents, and FAQs
• Access controls preserved on all connected documents
• Suggested prompts embedded in the UI for common advisor workflows
• Speed of deployment emphasized; flexible deployment into client-facing portals and advisor workstations
• Designed to automate standard functions, reduce calls to the home office, and support new advisor onboarding (100–150 per month)

Results

The deployment enabled 900 client advisors to access real-time, accurate answers to regulatory and operational questions through a conversational interface. The firm reduced onboarding friction for new advisors (100–150 per month) and reduced calls to the home office for standard procedure queries. The system supports compliance adherence by keeping advisors current with regulatory requirements. Quantitative efficiency metrics (e.g., time saved per query, reduction in compliance incidents) were not disclosed in the Squirro source.

Lessons

• RAG-based employee agents are particularly valuable in regulated industries where institutional knowledge is voluminous, frequently updated, and consequential if misapplied
• Access controls on underlying data sources must be preserved end-to-end — this was specifically noted as a differentiator vs. larger providers
• High-volume advisor onboarding (100–150 per month) creates a recurring use case for AI-assisted knowledge ramp-up that justifies ongoing investment
• The limiting factor is knowledge quality: the AI is only as good as the completeness and accuracy of the connected data sources