Eldad A. Fux, founder and CEO of Appwrite — one of the most actively maintained open source backend infrastructure projects globally with 55,100 GitHub stars and 800 contributors — argues in a March 2026 essay that AI tooling is reshaping who can contribute to open source. His thesis: the real barrier to contribution was never syntax fluency but the absorption of tribal knowledge scattered across old pull requests, review comments, and maintainer intuition. AI agents can now compress this onboarding curve by mapping file graphs, explaining subsystems, identifying naming conventions, and drafting tests that match a project's existing patterns — without requiring deep prior familiarity with the stack.
Fux writes from direct operational experience rather than theory. Managing 800-plus contributors across a complex microservices codebase covering authentication, databases, storage, serverless functions, and real-time infrastructure means he has personally observed the bottleneck he describes at scale. The shift he identifies moves the contribution threshold from "Do I already know this stack well enough?" to "Can I understand the goal, validate the result, and collaborate responsibly?" — a gate he characterizes as healthier for the ecosystem because it opens participation to domain experts who are not language specialists.
Projects that encode their philosophy explicitly through documentation, clear structure, and machine-readable artifacts will benefit most from AI-assisted contribution, because both humans and agents can produce aligned changes from the outset. Projects that rely on undocumented tribal knowledge will struggle. Fux also notes that forking becomes cheaper and more sustainable as implementation friction drops — which means the competitive advantage in open source no longer accrues to whoever built the most intricate implementation, but to whoever built the clearest one.
Fux's perspective is informed by his active development of Imagine, an AI-powered application builder launched by Appwrite in December 2025 that uses specialized agents to convert natural-language ideas into production-ready applications. He is not theorizing about AI-assisted contribution from the outside but building and operating agent infrastructure against the same open source codebase his essay discusses. The essay's core prescription — that maintainers must optimize for legibility, <a href="/news/2026-03-14-agile-manifesto-ai-addendum-prioritizing-shared-understanding-over-shipping">making implicit context explicit for both human contributors and agents</a> — reflects how the open source ecosystem will need to reorganize as AI-assisted contributions scale. For maintainers who have relied on complexity as a moat, that deadline is arriving faster than most have planned for.