2
1 Comment

I spent years wanting to build an app. AI coding tools helped me actually ship it.

I've worked in IT for a few years now. Decent job, stable income, and a quiet itch I could never scratch — I always wanted to build something of my own.

A few years ago I tried to learn Flutter. The plan was to build an app for both iOS and Android, cover both platforms, do it properly. I got a few weeks in before work priorities took over. The side project folder sat there. I told myself I'd come back to it.

The problem I kept living with
Thousands of photos on my phone, organised the way everyone's photos are "organised" — which is to say, not really. Camera roll going back years. Folders with vague names. And every time someone asks "do you have photos from that trip?" I spend ten minutes scrolling and still can't find them.
But the search problem wasn't what really bothered me. The photos felt like files. The memories felt like they were somewhere around the photos but not actually in them. The story of what happened, what I felt, who I was with, what made it special — lives only in my head. It faded.
There should be something that turns photos into an actual memory. Not an album. A story.

Why this time was different
I'd thought about building this for a long time. What stopped me wasn't the idea — it was the gap between having an idea and actually shipping something. That gap felt enormous when I was trying to learn Flutter from scratch while working full time.
Then AI coding tools got genuinely good. I decided to try again. React Native this time.

What I built
MemoBloom takes photos from your camera roll and turns them into a warm, narrated memory story — entirely on your device. No account. No cloud. Not a single photo leaves your phone.
The AI pipeline runs locally: a vision model describes the photos, then a language model writes a personal narrative from those descriptions. The result is an animated slideshow with a story you can actually read, search, and keep.
You can also search your memories by meaning — "rainy birthday dinner" or "Coorg monsoon" — and it finds the right one without you scrolling through thousands of photos.
And because the photos themselves are compressed into a memory, you can delete the originals and free up the storage. That turned out to be the feature people care about most. I didn't see that coming.

One honest caveat: The AI-generated narratives are warm but not perfect. Running a 0.6B language model on a phone means you're working within real constraints. The stories are good enough to be meaningful, but they're not what a skilled writer would produce. I'm iterating on the prompt engineering and model selection — it's getting better with each update.
The architecture is model-agnostic — the pipeline will benefit directly as on-device models improve. My hope is that within 12–18 months, a single multimodal model handles both the photo understanding and the narrative writing in one pass, which should improve quality significantly.

Where I am now
MemoBloom launched on App Store and Google Play recently.
Current numbers, honestly:

Downloads: somewhere between 11 and 50
Revenue: $0
Reviews: a few kind ones from people I know

That's it. That's the whole dashboard.
I know what the problem is — it's not the product, it's discovery. The App Store doesn't hand you users. I'm learning that part now. Reddit, IndieHackers, Show HN, short-form video. The build was the part I knew how to do. The distribution is the part I'm figuring out.

What I actually learned
AI coding tools didn't make the hard parts easy. The OOM crashes were still hard. The prompt engineering for the narrative quality took weeks of iteration. The App Store submission process has its own surprises.
What they did was compress the distance between "I have an idea" and "I have something running on my phone."

This week I'm trying Show HN, a few targeted Reddit posts, and building a habit on IndieHackers. If you've been through the "built it, now what?" phase of distribution — I'd genuinely love to hear what worked for you. Specifically curious if anyone's had success with App Store ASO or short-form video for consumer apps.
And if you want to try MemoBloom:

iOS: https://apps.apple.com/us/app/memobloom/id6766504864
Android: https://play.google.com/store/apps/details?id=com.minash.memobloom

Free for up to 10 memories. No account needed.

on May 31, 2026
  1. 1

    Ashwin, this is almost identical to where I am. IT background, wanted to build something for years, AI tools made the gap finally crossable.
    I'm building Anthilia — a mobile app for tracking recurring fixed expenses (subscriptions, utilities, loans) for the Italian market. Same stack, same phase: the build is done, distribution is the new problem.
    Your line about the App Store not handing you users is the thing nobody says out loud enough. I'm currently testing organic channels before committing to store submissions - waitlist first, validate the demand, then invest in infrastructure.
    On your question: I've found that the posts that work best are the ones that lead with the problem, not the product. "I kept forgetting about charges on my bank statement" lands differently than "I built a finance app." The product becomes interesting because the problem is recognizable.
    Curious how the semantic search feature is landing with users — that feels like the thing that's genuinely hard to explain in a screenshot.

Trending on Indie Hackers
Your build-in-public audience is not your market. I learned the difference the slow way. User Avatar 195 comments I built a WhatsApp AI bot for doctors in Peru — launched 3 weeks ago, 0 paying customers, and stuck waiting for Meta to approve my app User Avatar 62 comments Built a "stocks as football cards" thing. 5 days in, my launch tweet got 7 views. What am I missing? User Avatar 33 comments From broke and burned out as a PM, to launching my SaaS and optimizing my health User Avatar 32 comments Why Claude Skills Are Becoming Important for Tech Careers User Avatar 24 comments I kept starting projects and dropping them. So I built a system that wouldn’t let me User Avatar 23 comments