Scaling Agentic Wealth Management with AI-DLC
Event
About the event
Wealth management isn't one problem — it's dozens of specialist judgments happening at once: diversification, risk scoring, rebalancing, fundamentals, news sentiment, deep research. Cramming all of that into a single monolithic prompt doesn't scale. So what if each judgment had its own expert agent, and something intelligent orchestrated them on demand?
That's SmartLake: a fleet of independently deployable specialist agents running on AWS AgentCore, coordinated by a supervisor pattern, sharing a common plane of Bedrock models, per-user isolated memory, and an asynchronous SQS backbone. This talk walks through the full five-layer architecture — from client channels down to the data-and-platform layer — and unpacks the six dimensions that let it scale without friction: horizontal growth, concurrency, multi-tenancy, cost efficiency, resilience, and reuse.
But the real story is how it was built. Using the AI-DLC methodology (Inception → Construction → Operations), every capability was designed, generated, and deployed as a governed unit — with human approval gates at each stage and AI carrying accumulated context forward. The result: adding a new capability means adding an agent, not rewriting a monolith.
You'll leave understanding:
- Why composition beats monoliths for scaling agentic systems
- How to isolate memory per user safely as tenants multiply
- How AI-DLC keeps AI-generated code governed, reviewable, and production-ready
- A concrete, reusable pattern for shipping agent fleets on AWS
Who should attend: Architects, engineering leaders, and builders working with LLM agents, AWS Bedrock/AgentCore, or fintech platforms.