Rethinking Workflows in the Agentic Era

Ready to move beyond simple chatbots? Learn a practical methodology for building multi-agent systems that automate high-value, complex workflows like code migration and test generation.

#1about 5 minutes

Moving beyond chatbots to complex AI-driven workflows

Common single-shot AI use cases are limited, so the focus should shift to automating complex, multi-step processes that mirror human workflows.

#2about 4 minutes

Selecting and evaluating workflows for AI automation

Choose non-time-critical processes where accuracy is more important than speed and the output can be scored using objective, verifiable evaluation criteria.

#3about 3 minutes

Decomposing human processes into agent-driven units

Map an existing workflow by identifying actors and decision points, then break it down into goal-driven, injectable units with clear input-output contracts for each agent.

#4about 4 minutes

Designing and implementing multi-agent collaboration patterns

Structure agent interactions using patterns like linear chains, branching, or dynamic orchestration, and scale the system by refining prompts and optimizing model choices.

#5about 7 minutes

Demo of an agentic system for Azure architecture design

A multi-agent system uses an Azure expert, a cost specialist, and a project manager to design cloud architecture and incorporates a human-in-the-loop via a phone call for feedback.

#6about 4 minutes

Demo of automating unit test generation and validation

A workflow automates unit test creation by using separate agents to generate pseudo-code, write C# tests, and then compile and run them to validate code coverage.