Development Workflow with AI

Development Workflow with AI

How should business application developers work in this agentic age? At the highest level, the tasks remain the same: understand the business rules, create clear requirements documents, and iterate on code to build strong abstractions. “Garbage-in, garbage-out” still holds true. If we accept that we now have tools to handle the tedious, time-consuming jobs of writing code and creating tests, we should use that freed-up time to step up as Senior Engineers and System Architects.

The book Learning Systems Thinking by Diana Montalion details how software engineering is just one part of an organization’s entire system. This post is the beginning of an exploration of my transition from a software developer to a systems thinker and systems architect. There will be more on Systems Thinking in later posts. For now I am trying to learn how to develop applications

Sequece Diagram of Development Workflow

sequenceDiagram
    autonumber
    actor Dev as Developer
    actor Agent as Agent
    participant Jira as Jira
    participant Confluence as Confluence
    participant GitHub as GitHub
    participant Apps as Apps

    rect rgb(240, 248, 255)
        Note over Dev, GitHub: Phase 1 - Planning and handoff
        Dev->>Jira: Create or review the work item
        Dev->>Confluence: Capture business rules and requirements
        Dev->>GitHub: Open or link the implementation branch / PR
        Dev->>Agent: Hand off the task context
    end

    rect rgb(245, 240, 255)
        Note over Agent, Apps: Phase 2 - Implementation
        Agent->>Jira: Read acceptance criteria and status
        Agent->>Confluence: Pull business rules and documentation
        Agent->>GitHub: Inspect repository, branch, and PR context
        Agent->>Apps: Develop code for the requested change
        Apps-->>Agent: Return implementation output and test results
        Agent->>GitHub: Commit changes and push updates
        Agent->>Jira: Update work item progress and notes
    end