Agent Driven SDLC
Overview
Agent Driven SDLC is a two-day, hands-on course for software teams adopting coding agents across delivery work. Rather than presenting an agent as a prescribed replacement for developers, the course treats it as a configurable collaborator. Attendees decide where delegation helps, where deterministic automation remains better, and where human judgment must stay in control. Short demos and labs build a delivery model that works across IDE assistants, terminal agents, internal platforms, and continuous integration services.
Day one connects product intent, architecture, repository context, task definition, and implementation. Participants convert an ambiguous request into testable outcomes, concise constraints, useful examples, and explicit non-goals. They prepare a project so people and agents can find its code, commands, interfaces, decisions, and safety rules. Additionally, they assess context sources, sandbox boundaries, data handling, and permission models before agents receive tools. Comparing interactive sessions, pull-request automation, and event-driven workflows helps teams match autonomy to risk.
Day two makes verification and delivery central to the design. Students use acceptance checks, code review, system tests, security checks, traceability, and rollback planning to evaluate candidate changes. However, the course distinguishes evidence suited to a small reversible fix from assurance required for a high-impact release. Attendees then explore review-agent, paired-agent, and multi-agent patterns alongside simpler single-agent workflows. Each design includes ownership, handoff, integration, cost, and failure-mode tradeoffs.
The course closes with a delivery-lifecycle design exercise. Teams select an operating model for a realistic product change, define its human control points, and create a measurable improvement plan. The result is not a fixed agent playbook. Instead, participants leave with a framework for evaluating new tools and practices as capabilities, regulations, and team needs change.
Duration
2 Days
Who Should Take This Course
Prerequisites
Participants should have experience developing software in a team and using Git-based source control. They should understand the basics of building, testing, reviewing, and releasing software. Familiarity with a coding assistant is helpful but not required. The labs use a small application, so attendees should be comfortable reading code and running common development commands.
Why You Should Take This Course
Upon completing Agent Driven SDLC, attendees will be able to:
- Assess where coding agents, conventional automation, and human decision-making fit within a software delivery lifecycle
- Turn an ambiguous change request into a bounded, testable agent task with acceptance checks, constraints, and non-goals
- Prepare repository context, reproducible environments, and delivery guidance that support reliable agent work
- Define appropriate data, tool, credential, network, and approval boundaries for different classes of work Direct an agent through investigation, planning, implementation, review, and correction without
- surrendering engineering ownership
- Apply tests, static analysis, security checks, independent review, and release criteria as evidence for agent-assisted changes
- Compare interactive, pull-request, continuous integration, single-agent, and multi-agent operating models against risk and coordination cost
- Design measurements, governance, and feedback loops that improve agent-enabled delivery over time
Course Outline
Agent Driven SDLC
- The Changing Software Delivery Lifecycle
- Building an Agent-Ready Engineering Foundation
- From Product Intent to Executable Work
- Engineering Assurance for Agent-Assisted Changes
- Choosing an Agent Operating Model
- Collaboration and Scaled Agent Work
- Delivery, Release, and Operations
- Evaluation, Governance, and Continuous Improvement