Building BeeBuzz: A Privacy-First Experiment in the Age of AI

BeeBuzz is a small project that I’m building in my spare time. There’s no VC behind it, no growth roadmap, no team pushing features every week. It started from a very practical need.

I use Home Assistant for my automations, and I needed a reliable way to send push notifications. The first step was obvious: look for existing open source solutions. I strongly believe in open source, and my default approach is always to build on top of what already exists.

While exploring different options, I noticed most of them either offered far more features than I needed, required mobile apps, or were built around a different model. I didn’t need a full notification platform. I just wanted to send a secure push notification when an event happens. Even in self-hosted mode, I wanted messages and attachments to remain protected in case the server is compromised. No software is 100% secure, so the architecture should assume that risk. That requirement changed everything.

BeeBuzz is intentionally web-only. I didn’t want to build Android or iOS apps because that would have introduced a completely different level of complexity: release cycles, store policies, background services, and constant maintenance. By staying on the web, I can keep the surface area small and focus on simplicity, security, and clarity.

This is also an experiment in building software in the age of AI. I used AI during the early stages to prototype ideas and validate architectural feasibility. It also helped me generate boilerplate, scaffold components, and even draft tests faster. But nothing goes into production unless I write it, review it, and fully understand it. AI supports the thinking and coding process — it doesn’t replace responsibility.

Working on BeeBuzz outside of my day job shapes many decisions. Progress is steady but not fast, and that’s intentional. There are no artificial deadlines and no pressure to add features just for growth. I optimize for clarity, maintainability, and long-term sustainability. Keeping the project independent gives me that freedom.

When I say privacy-first, I don’t mean it as a slogan. It shows up in architectural constraints: minimal readable metadata, optional end-to-end encryption, attachment TTLs, and a clear separation between trusted and encrypted modes. Some systems try to solve every possible use case. BeeBuzz is intentionally narrower. It aims to do one thing well.

The project will be released as open source. You’ll be able to read the code, self-host it, and inspect every decision. There will also be a SaaS version. The SaaS makes the project sustainable — and if you don’t want to self-host, it’s there for you too.

You can sign up for early beta access on the landing page.

I’m writing this devlog to document the reasoning behind the project. Not because the choices are perfect, but because they are intentional. Following the philosophy I use for my open-source projects, you can expect posts to appear on a practical, ongoing basis — not on a fixed schedule, but consistently enough to track decisions and progress. If you want deeper technical details, the source code will always be there.

BeeBuzz is small, independent, and web-only. It started from one real problem — a push notification that had to be private even if the server wasn’t — and every decision since has followed from that.