How To Build a GenAI-Augmented Software Organization • Marko Klemetti & Kris Jenkins • GOTO 2025
Marko Klemetti and Kris Jenkins explore how Generative AI is not just another tool but the next major paradigm shift in software development, forcing a radical reinvention of team structures, management roles, and the entire SDLC to achieve a 'five-minute-to-production' North Star.
GenAI: The Third Paradigm Shift in Software Development
Following the Agile and DevOps movements, Generative AI represents the next significant paradigm shift in the software development life cycle (SDLC). The ultimate “North Star” for a GenAI-augmented organization is the ability to take a complete feature from concept to production in just five minutes. This ambitious goal, reminiscent of the early days of continuous integration where a developer could get a green or red build in the time it takes to grab a coffee, is now becoming feasible for entire features.
Tools like Lovable and GitHub’s co-pilot and coding agents demonstrate this capability by using conversational UIs to generate and deploy features on existing or new codebases. This isn’t about incremental improvement; it’s a fundamental change in how digital solutions are produced.
Organizational Restructuring is Non-Negotiable
A common but flawed approach for organizations is to use GenAI simply to improve the efficiency of their existing structure. Asking, “Can we produce the same amount of code with fewer people?” is the wrong question. This leads to sub-optimization. For instance, if R&D budgets only allocate 5-10% to actual software development, even a 50% efficiency gain only improves the overall R&D output by 2.5-5%.
The real challenge lies in Conway’s Law: if the organizational structure remains the same, so will the outputs, regardless of the tools used. The true potential of GenAI is unlocked not by fine-tuning the existing workflow but by completely redesigning the organization to eliminate handovers and silos.
A Blueprint for the Future: Team Topologies and Platform Engineering
The “Team Topologies” model provides a valuable framework for this restructuring. The core idea is to orient teams around customer value:
- Stream-aligned teams: These teams are focused directly on delivering features and value to the end-user.
- Supporting teams: Other teams exist to support this value stream, including:
- Platform teams: They provide an internal developer platform that reduces cognitive load and enables stream-aligned teams to work efficiently without getting bogged down in configuring tooling.
- Enabling teams: They help stream-aligned teams adopt new skills and practices.
- Complicated subsystem teams: They manage complex components (e.g., a specialized database service) that the stream-aligned teams consume.
Platform engineering is the practical implementation of this, creating a unified environment for development, from version control and CI/CD pipelines to templates and knowledge bases.
The Rise of AI Agents
In the future, AI agents will be managed and orchestrated through these platforms. Instead of developers building bespoke solutions for common problems, they will leverage pre-built, platform-provided agents. For example, a developer could apply a “GDPR compliance agent” to their pipeline to automatically check every change, with the platform team being responsible for maintaining and updating the agent itself.
The Real Disruption: The Future of Management
While many fear GenAI will replace developers, the more significant impact may be on middle management. The goal is to move towards flatter organizations composed of small, self-organizing, “pizza-sized” teams that can deliver features independently.
In this model, the role of management shifts from overseeing tasks and people to:
- Setting the vision: Defining the direction and goals for the teams.
- Managing experimentation: Fostering a culture where teams can experiment, measure results, and iterate quickly.
This requires a fundamental change in management structure, particularly in traditionally hierarchical organizations. The knowledge has always resided with the people creating the solutions; a flatter structure empowers them to act on it directly.
Who Will Win and Who Will Fail?
The path to success with GenAI is not straightforward, and two distinct patterns are emerging:
- The Path to Failure: Organizations that simply use AI to cut costs and reduce headcount without changing their processes will fail. They will get stuck in a cycle of sub-optimization, and the increased feature velocity will likely lead to “feature creep,” degrading the end-user experience.
- The Path to Success: The winners will be those who reinvent their entire way of working. This is often easier for new companies that can start from scratch. For established enterprises, a successful strategy is to create a separate “skunkworks” entity, as Volkswagen did with its Cariad division. They tasked a new organization with building the software for their electric vehicles, creating a new “playbook” for software development that could then be applied back to the parent company.
Ultimately, the most successful organizations will be those that can create new solutions and business models at a fraction of the cost and time of their established competitors.