The Minimum Viable Unit of Saleable Software (brandur.org)

In a reflective blog post, software developer Blake discusses his shift from a stable role to building River, a job queue project, amid the evolving landscape of AI-driven software development. The rise of large language models (LLMs) has prompted questions about the viability of traditional software companies as businesses face the possibility of LLMs rendering their solutions obsolete. While LLMs have reduced development costs, the author highlights that even with their assistance, building complex systems requires continuous human oversight, adjustment, and maintenance, making a DIY approach less straightforward than it may seem. Blake argues that a new calculus has emerged in the "buy vs. build" debate, with certain software falling into a "zone of viability." This zone consists of products where the costs of creating customized solutions with LLMs surpass the licensing fees for existing software. He uses examples of Jira and Salesforce to illustrate that while lower-priced tools may not justify a rebuild, high-cost software like Salesforce could encourage in-house development. As he positions River within this framework and touts its unique offerings and competitive pricing, Blake implicitly invites the AI/ML community to consider how LLMs have reshaped the software development landscape and the implications for emerging software businesses.