About
I started Ashfall because I kept building tools for myself and realizing they solved problems other people had too.
The pattern was always the same. I'd hit some friction in my own life, whether that was tracking niche content across a dozen sources, or planning meals around my daughter's food allergies, or monitoring software releases across 50 projects in my homelab. I'd look at the available solutions and they'd be either too generic, too expensive, or pointed at someone else's use case entirely. So I'd build the thing.
After doing this enough times, I noticed the infrastructure underneath was starting to repeat. The same data acquisition patterns, the same classification pipelines, the same user-management plumbing. Ashfall is the decision to stop treating that shared infrastructure as throwaway and start treating it as a platform.
What I believe
The tools you use should work for you. Not for an algorithm, not for an ad platform, not for a growth team optimizing engagement at the expense of your attention. If a product can't articulate who it serves without using the word "everyone," it probably doesn't serve anyone particularly well.
Owning your infrastructure matters. Not as a flex. As a discipline. When you understand every layer of the stack from the hardware to the application, you make better decisions about tradeoffs. You don't build features because a vendor's pricing model incentivizes them. You build what actually helps.
The things people care about deserve serious tools. I don't think it's useful to call them "hobbies." When someone spends years building a collection, learning a craft, developing expertise in a niche that most people don't understand, that's not casual. That's a real part of their life. The tools that serve those interests should take them as seriously as the people who have them.
AI is an implementation detail, not a feature. Language models are genuinely useful for classification, extraction, and structuring unstructured data. They're embarrassing as a marketing category. A product should be useful enough to stand on its own. If the only interesting thing about it is that it uses AI, it probably isn't interesting.
What I don't do
- Engagement optimization. I'm not trying to keep you on a page longer.
- Data brokering. Your information stays yours.
- Growth-at-all-costs. Ashfall is self-funded and plans to stay that way.
- "AI-powered" marketing. The word AI does not appear in any product's tagline.
How I build
I'm a solo founder with 20+ years of engineering experience. I work pseudonymously by choice. The work speaks, and I'd rather people evaluate the products than the person behind them.
Ashfall Systems is the trade name under which the products ship — the legal entity behind it is Kota Ridge LLC, which also holds the infrastructure everything runs on.
The technical foundation is Elixir/Phoenix for application logic, PostgreSQL for data, Dagster for orchestration, and a self-owned Kubernetes cluster for everything that runs. The infrastructure exists because I believe in understanding the full stack, and because running your own compute turns out to be significantly cheaper than cloud when you're doing GPU inference at any real volume.
Every product I ship under Ashfall shares the same platform layer: accounts, billing, email, event logging, and marketing tooling. When I build something new, the foundation is already there. That's the whole point. The infrastructure is the asset. Products come and go. The platform compounds.
— Erik, Ashfall