Developer Coffee Chat: Building a Simple Risk Assessment Agentic Workflow

In this Coffee Chat, we’ll build a loan risk assessment workflow that looks complex on paper (docs, extraction, LLM scoring, routing, human review) but stays clean and simple because it includes sub-workflows.

We’ll walk through how to split the workflow into small, reusable pieces (like a sub-workflow for document validation + classification), then plug that into a main flow that runs LLM-based risk scoring and routes uncertain cases to a person for approval. The goal is a workflow that’s easy to extend, easy to debug, and easy to explain to your team.

No prior Conductor experience needed. We’ll keep it approachable, explain the moving parts as we go, and focus on the patterns you can reuse in your own workflows. Questions about agentic workflows are welcome, if this is something you are interested in learning more about.