As we move forward at light speed toward the implementation of Agentic AI and Artificial General Intelligence (AGI), I think it’s time to consider a next-generation Software Development Life Cycle in terms of team structure.
So, what is Agentic AI?
Agentic AI refers to AI systems that operate with a level of autonomy, decision-making, and adaptability similar to human agents. These AI models can independently plan, execute tasks, and adjust their behavior based on goals, feedback, and environmental changes.
Since there is no real Agentic AI on the market at the moment, we need to consider how to incorporate these systems into current or future IT teams.
I would assume that, in the near future, there will be several proposals for AI-based developers—either universal or language/technology-specific. As a leader, you will need to compose a hybrid team of both human IT specialists and Agentic AI virtual specialists.
Another assumption is that there will be more than one proposal on the market, meaning we’ll likely see different AI agents (on-premises as LLM/RAG, online, in private/public clouds, etc.). One of the most important questions here is security. I would classify the key parameters as follows:
- Security
- Cost
- Performance
- Redundancy
- Special Requirements (content window, fine-tuning, integrations, etc.)
If you’re working in government or finance, you should start thinking about on-premises/private cloud resources in advance. For any other commercial use of agents, you can begin evaluating options for the level of security your business model can tolerate.
I believe that hybrid teams are the near future. For example, in a development team, you might have a set of Agentic AI developers and human dev/engineering leads with specialized AI-related skills. It’s likely that the role of a regular programmer or coder will diminish quickly.
One of the interesting new fields is Agentic AI synchronization and communication interfaces. Such interfaces should be secure, fast, and agnostic to different model providers.
I also believe that a “next-gen Jira” could be integrated with these interfaces, allowing a dev lead to assign tasks directly to AI models or let the models select tasks themselves, much like in a classic Agile environment.
Finally, I see that the positions of Scrum Masters, Performance Engineers, and QA specialists could also be handled by Agentic AI. My hope is that software quality will rise as new approaches to automated testing are implemented.
Summary:
I believe that we are (surprise but again) pretty close to the transformation point, and it is wise to start planning the organizational transformation in advance and understand how to incorporate Agentic AI (or AGI) into the existing (or new) IT teams as frictionlessly as possible.
Top comments (0)