I built Aural because most mock interview tools felt like forms with AI bolted on. I wanted the experience to feel closer to an actual conversation, with spoken answers and follow-up questions based on what you say. This month, I hit a milestone that mattered more to me than another feature launch: I released the full product under the MIT license.
The biggest lesson was that open sourcing a SaaS exposes every shortcut. “Works on my deployment” is not enough. Authentication, database permissions, environment variables, model configuration, and setup docs all become part of the product. Preparing the release forced me to simplify the architecture and treat self-hosting as a real user journey, not a checkbox.
Aural is now 100% TypeScript, built with Next.js, Supabase, and PostgreSQL. It supports pluggable LLM APIs and uses Row-Level Security for data isolation. The hosted version has live demos without signup, plus a free tier with 3 templates and 100 AI tokens. You can try it at aural-ai.com or inspect the code on GitHub. For founders who open-sourced a SaaS, did it help distribution or mainly increase support work?