An AI DevOps assistant is a combination of the Project Assistant, Business Administrator (BA) Assistant, Solution Architect (SA), Scrum Agent (Project Assistant), Developer (Dev) Assistant and the Quality Assurance (QA) Assistant.
Assigning work packages to a DevOps AI Assistant ensures your templates are used to:
- Business Analyst
- Document Business Requirements (BA Assistant)
- Gets approval of Business Requirements (workflow to user)
- Solution Architect
- Document Technical Specifications (Dev Assistant)
- Get approval for Technical Solution (workflow to user)
- Scrum Manager / Product Owner
- Prioritizes work packages (workflow to user)
- Developer
- Development is executed and tested against the Technical Specifications
- Quality Assurance
- Quality Assurance (QA) prepares detailed test scripts to valid the business requirements
- Test Completion Matrix is created
- Tests are executed and resolved with the Developer; blockers are highlighted to the Scrum Manager
- End Users
- Test the new features in a Test Environment
Upon user acceptance of the work package, all documentation is packaged for release to production.
A separate AI DevOps can be initiated for each work package; this ensures a tighter context from requirements to solution.
I don’t want to underestimate the complexity of this agent flow; packages like CursorAI will accomplish 75-90% of these tasks if prompted correctly. The complexities arise around creating and refining AI prompts and access to, and training of, the production codebase.
Using existing agents and models is quick and would address most development.
An inhouse solution is more involved:
- Setting up an AI environment (cloud or onsite)
- Integrating with Large Language Models (LLM) and Coding trained models
- Training your model on your documentation, testing, development and process procedures
- Training your model on the production codebase
- …
If you haven’t started, you’re probably already behind.