AI coding agents may be about to revive the practical relevance of free software. When an agent can read, understand, and modify source code on a user's behalf, the difference between open and proprietary software becomes concrete rather than philosophical. A personal attempt to build a tweet-to-Sunsama workflow illustrates the problem: six layers of workarounds were needed because Sunsama has no API and iOS is closed. With free software, an agent could have solved the same problem in minutes. The post traces the history from Stallman's four freedoms through the 'open source' rebrand and the SaaS loophole that made software freedom feel irrelevant, then argues agents flip the equation by acting as technical proxies for non-technical users. Counterpoints are acknowledged: self-hosting costs, open-source sustainability risks from AI-generated contributions, and the collapse of the user-maintainer feedback loop. The conclusion is that 'can my agent fully customize this?' will become a mainstream software evaluation criterion, and SaaS products without genuine moats face serious risk as agents collapse switching costs.
Table of contents
Free software once mattered deeply, but faded when SaaS made those freedoms feel irrelevant.The “open source” rebrand preserved code sharing while stripping out the user-rights philosophy.SaaS scaled by exploiting a licensing loophole that let vendors avoid sharing their modifications.My Sunsama workflow shows how closed SaaS blocks useful customization even for motivated users.I tried to build the workflow anyway, and every closed layer turned a simple idea into a brittle hack.AI agents can finally exercise software freedom on behalf of people who cannot code.This shift is already visible across multiple thinkers who see agents increasing the value of openness.The pendulum can swing back toward openness, but convenience and maintainer economics still matter.The next buying criterion will be whether your agent can actually change the software to fit your needs.Sort: