In the ever-evolving landscape of software development, the concept of multi-tenancy has undergone a profound transformation. Once defined by the needs of human developers, it is now being reshaped by the capabilities and demands of coding agents. This shift is not merely a technical adjustment; it's a paradigm shift that challenges long-held assumptions about how we structure and manage development environments. The traditional view of multi-tenancy, where the tenant was the developer or a group of developers, is being replaced by a new model where the tenant is the change itself. This new perspective is particularly fascinating because it directly addresses the challenges posed by the rapid advancement of AI and automation in software development. In this article, I will delve into this transformation, explore its implications, and discuss how it can be leveraged to create more efficient and scalable development platforms. The traditional multi-tenancy model, which has been in place for 60 years, has been characterized by the tenant being the organization or a specific team within it. This model worked well when changes were introduced at a human pace, with each developer contributing one change at a time. However, with the advent of coding agents, this model is no longer sufficient. Coding agents, such as those used by Anthropic and Cursor, can run multiple sessions simultaneously, each with its own changes in flight. This has led to a significant increase in the number of changes being made, with Microsoft's study finding a 24% increase in pull request merges over four months. This surge in changes has implications for capacity planning and resource management. The traditional approach, which plans capacity based on the number of developers, is no longer effective. Instead, we need to think in terms of changes in flight, which can number in the hundreds for a 50-developer organization. This shift in perspective is crucial, as it highlights the need for a new model of multi-tenancy that is change-centric, rather than person-centric. The new tenant is not the developer or the agent; it is the change. This change-level tenancy has three key requirements: near-free creation, isolation that covers only what has changed, and a lifecycle bound to the change itself. These requirements are not new, but they have not been applied to the platform that builds the product. In fact, platform teams already run this playbook in production, where tenants share the substrate, each tenant owns only what makes it distinct, and creation is self-service and cheap. The key difference is that this playbook is now being applied to the platform that builds the product, rather than just the product itself. This shift in focus has significant implications for the way we design and manage development platforms. By treating every change as a tenant, we can create a more efficient and scalable platform that is better suited to the demands of modern software development. The new model of multi-tenancy, where the tenant is the change, has several advantages. Firstly, it makes tenant creation nearly free, as isolated data no longer requires copying a database. Instead, Neon and Xata create copy-on-write branches in seconds, consuming storage only for the data that diverges. Secondly, it enables automatic offboarding, which keeps the accounting accurate and prevents the accumulation of orphaned environments. This is particularly important at the change scale, where orphans can accumulate as fast as agents abandon experiments. Finally, the new model supports the development of an agent-native software development lifecycle (SDLC), where the change is the only unit of isolation that stays stable when the workers become software. This shift in perspective is crucial, as it allows organizations to leverage the power of coding agents without being constrained by the limitations of the traditional multi-tenancy model. In conclusion, the transformation of multi-tenancy from a person-centric to a change-centric model is a significant development in the field of software engineering. It challenges long-held assumptions and opens up new possibilities for creating more efficient and scalable development platforms. As we continue to embrace the capabilities of AI and automation, this new model will become increasingly important, enabling organizations to stay ahead of the curve and deliver innovative solutions to their customers.
