What happens when the reinvention triggered by AI forces GSIs to rethink the very core of their operations — the (now infamous) SDLC?
Something deeper than faster coding is happening in software engineering. AI is blurring the boundaries of the traditional Software Development Lifecycle and pushing it toward what can only be described as an Agentic Development Lifecycle (ADLC) — where intelligent agents participate across the lifecycle rather than sitting beside it as coding assistants.
The core lifecycle is becoming less linear and far more continuous, adaptive, and orchestrated. When agents begin to decompose requirements, generate architectures, write code, test systems, and monitor production — the roles across the entire engineering ecosystem begin to evolve. We are already seeing the emergence of new roles.
Business Prompt Engineers
A new generation of professionals who focus less on writing detailed specifications and more on defining intent, outcomes, and constraints — guiding AI systems to translate business needs into executable system designs.
AI Systems Architects
Developers transitioning to this role are focused on:
- System architecture and design intent
- Orchestration of multiple AI agents
- Domain and product understanding
- Governance, safety, and reliability of autonomous systems
Specialist AI Assurance Engineers
Taking up responsibilities for validating system resilience, adversarial behaviour, security vulnerabilities, and trust boundaries. As AI begins writing more of the software, the emphasis moves from functional testing to trust, safety, and cyber resilience.
AgentOps Engineers
Professionals who manage AI-driven delivery systems, monitor agent behaviour, maintain observability, and design rollback and governance mechanisms when autonomous systems make decisions.
This is why the current wave of AI discussions in the industry feels so intense — it is not just about tools. It touches the very foundations of software organisations — roles that have existed for decades and processes that have been refined over generations.
In some ways, it feels like AI has undone the last 30 years and handed us a fresh starting point. And that is exactly why we must unlearn, reinvent, learn again, and move forward. Organisations must not just build software faster. They must learn to build software differently.