Financial Services & Fintech · Agentic Intent Capture
Agentic Intent Capture: AI Agent on Bedrock
Supporting the cloud foundation behind regulated lending conversations and handoffs
The Client · The leading US digital mortgage lender
Overview
The leading US online mortgage lender engaged Taller in 2025 to support AI-capable and cloud-infrastructure engineering across its AWS and Bedrock stack. Taller’s work underpinned the lender’s agentic AI platform.
The Problem
The lender’s AI Agent strategy centered on a single AI Agent API that routed requests to eight specialized AI agents, built on Amazon Web Services' Bedrock, Amazon’s managed service for building generative-AI applications, covering origination, servicing, and broker support. Each agent had its own knowledge base, memory, defined actions, and guardrails, and handed decisions back to a human banker whenever a regulated lender could not delegate the call. The architecture was the lender’s own; it required AI-capable engineering to operate at scale within strict regulated-lending safeguards for security and threat modeling.
The Solution
Taller contributed AI-relevant individual-contributor roles and cloud-infrastructure engineering to the lender’s AWS platform.
The Impact
These results belonged to the lender’s eight-agent architecture and guardrail discipline; Taller’s contribution was the engineering capacity behind it.
more likely to close a loan after engaging the AI Agent
expected improvement in lead conversion
