Why I'm Shutting Down Dear Self

I love journaling. I've kept one, in some form, for years. But I kept running into the same three walls with a physical notebook. I couldn't search it. I couldn't attach anything to it. And, the thing that bothered me most, it never gave anything back. I'd fill page after page and never notice the patterns sitting right there in my own handwriting.

So I did what a lot of us are doing right now: I decided to learn how to actually build agentic AI applications, not just talk about them. Journaling was the perfect excuse. I wanted something that read what I wrote, remembered it, and told me what it noticed — a mood creeping back, a topic I kept circling, a question I kept avoiding. That project became Dear Self. You emailed it like you'd email a friend, and it emailed you back.

It worked. People used it. I'm proud of it.

Here's the part I have to be honest about: the infrastructure behind it is expensive, and it does not get cheaper as more people use it. Every entry gets processed, embedded, and reasoned over by models that cost real money per call. Running that for myself is one thing. Running it for everybody, for free or near-free, is not something I can sustain. I looked at the numbers for a while and kept hoping they'd look different. They didn't.

So I'm killing the public version.

As of today, you can't email Dear Self and get a reply anymore. If you want it, the code is open source, and you can deploy it on your own infrastructure and use your own API keys: github.com/dearselfai/application. If you were a user and want your data exported or deleted, email me directly at yordan@dearself.ai.

I don't regret building it. I regret that it can't stay open to everyone. But I'd rather say that plainly than quietly let the inbox go dark.