Now what?

Hello. I’ve been back in college for a little over a month now and have not missed SF or NYC at all, which has surprised me.

The first two days I was back here, I was complaining that I was back in the Matrix and college was boring. Then, I decided to do things other than just building and started having fun again.

So far, I’ve:

• gone to Chinese Calligraphy club

• played Intramural volleyball with my friends

• started getting into pickleball and playing multiple times a week

• went to my first college football game (we lost, of course)

• went to Chicago for Labor Day weekend with my friends and stole food from Trump Tower

So yeah. I’ve been really trying to do things that I can only really do in college because I might as well make the most of it. Maybe I’m back in the Matrix, who knows.

Of course, I haven’t stopped building. The idea I talked about in my last time was a semantic blocker for Twitter. Almost like prompting your feed. I launched twice and no one really cared and then I realized I didn’t really care so I decided to move on. I had fun making a banger launch video and the landing page so that was cool.

For the past two weeks, I’ve been working on a new idea and unlike the rest of my graveyard of half-baked projects, I’m not in a rush to launch it and see how much traction I get. I’m building an AI journaling app. Actually, I hate calling it an AI journaling app, but saying “a journaling app with AI” sounds kind of weird. But I think of it that way because the journaling part is the main focus. The AI happens to be a little side character that’s there when you need it.

“Okay, so what does this AI journaling app actually do.”

Doesn’t do much right now. It just looks like a text editor right now and that’s what it pretty much is. You set the amount of time you want to write for and then you just write for that long.

But at the end, you see a little “Go Deeper” button that allows the AI to ask you an introspective question. The AI can also look through your past journal entries to ask you better questions, which is pretty cool.

I recently switched out the model to Claude 4 Sonnet and it just dramatically improved. It just understood when to look for context versus way better. For instance, this banger response:

I’ve had a lot of fun playing around with the UX of this. Initially I had it inline but didn’t like how I could always see the AI responses. Then I switched to a more google docs style thing where I could see the AI response as a comment on a right sidebar but then my eyes would have to switch and all that. So I decided to combine both approaches to have inline responses, but I could toggle them to be shown or hidden (default hidden) and that seems to be fun.

The thing that’s different about this is that regardless of me building this, I journal. It’s a habit I’ve been doing pretty regularly for at least five years. Before building this app, I had a long Apple note of almost 4 months of entries and it was getting pretty slow. My goal is to build something that even if no one sees or I don’t post about it, I still keep using it because I just…like the product. Kind of paradoxical but I feel that if I’ve done that, there’s definitely something with this idea. And I’m in no rush to make another banger launch video because I just want to make the product really damn good.

This is, of course, the opposite to all the startup advice that’s out there. “Launch fast, iterate, don’t be building for too long.” I feel that that advice doesn’t necessarily apply when YOU are the user of the product.

Also, onboarding a close friend this Sunday at 3 pm, so I can set a hard deadline for myself and get someone else dogfooding this product with me as well.

So yeah, that’s what I’ve been up to. Reach out anytime and if you want to be onboarded later on, let me know.

Till next time.

- Sahil