Free Lesson
Design a Research Agent with Decision Models
Part of Build a Research Agent: Decisions, Workflows and Tools
30 min
Nov 12, 2026 11:00 AM
What you'll learn
Choose the next research action with a decision model
Configure runtime actions so the model selects a tool or asks for context instead of generating a long answer.
Connect the agent to company reports and Python
Retrieve two issuer reports and compare reported revenue with earlier guidance using a Python calculation.
Turn an observation into a testable research question
Inspect the evidence record and a failure case, then define what to test across other earnings announcements.
Why this topic matters
Design an agent that chooses what to investigate, selects a tool, and uses the result to decide what to do next. We will connect an open decision model, document retrieval, and Python in an earnings-research example. Inspect the action choices, evidence record, and a failure case, then turn the observation into a testable research question. No trading background is required; basic Python will help you adapt the example.
