Trying to build paid offline AI tools on iOS without subscriptions – is this a bad business idea?

Hey Indie Hackers 👋

Last month I posted here about a tiny suite of offline AI tools I’m building under the name FanStudio – everything runs on your device, no account, no tracking.

Quick recap:

I’m David, solo iOS dev
Building small offline AI utilities for iPhone / iPad
Two apps live right now:
- Photo Space Saver – helps people deal with the “iPhone Storage Almost Full” nightmare
- DocSnap & Summarize – scan, OCR, and summarize docs completely offline

My self-imposed rules:

No accounts
No analytics SDKs
No servers touching user data
Prefer one-time unlocks, not subscriptions

On the “human” side this all feels great.
On the “is this a real business?” side… I’m still not sure 😅

So I wanted to share what I’ve learned so far and get some advice from people who’ve been doing this longer.

Why I’m so stubborn about offline

There are a few categories of data I really don’t like sending to random servers:

Personal photos and videos
Financial / legal stuff
Anything that includes other people’s info

Most AI apps default to: “upload everything to our backend, trust us”.

I basically flipped it to:

“If it can’t run on-device, I don’t ship it.”

That means:

No “just call an API” shortcuts
Lots of time spent squeezing models into small binaries
Accepting “good enough offline AI” instead of “huge cloud models that can do everything”

Users seem to genuinely like the offline / no cloud angle.
It makes the trust conversation a lot easier.

And as a solo dev, it’s nice that I don’t have to worry about GPU bills or infra blowing up on launch day.

Where it starts to hurt

The part I didn’t think through enough is what this does to the business side.

1. No analytics = slower feedback

Because I avoid analytics SDKs, I don’t really know:

Where people drop off in onboarding
Which screens/features are used most
What pricing experiments actually work

Right now my “analytics” are:

App Store reviews
Emails people send me
Occasional long feedback messages (which are gold)

It feels clean, but honestly it’s also like driving at night with the headlights on low beam.

I’m still trying to find the line between:

respecting privacy
and having enough data to not make dumb decisions

2. On-device AI has real limits

Cloud models make you lazy. On-device models constantly remind you of physics.

I’m always juggling:

Model size vs app size
Accuracy vs speed vs battery life
Old devices that technically support the OS, but struggle in real use

For example:

DocSnap ideally would handle every kind of document in every language
In practice I have to pick my battles: which languages, which use cases, what I can pack into a reasonable app bundle

There are moments where I think:

“If I just had a small backend, this would be so much easier…”

…and then I remember I’m the one who said “no backend” in the first place 😅

3. One-time pricing in a subscription world

Personally I hate being subscribed to 20 different tools.
So for now both apps are:

Free download
Limited free use
One-time unlock for full features

The good:

Easy to explain
Aligns with the “offline, low-stress” vibe

The hard part:

Revenue is spiky: launch → small bump → flat
There’s no built-in recurring piece
I have to ask myself “how much ongoing work can I justify per app?”

I know there are ways around this (paid upgrades, bundles, pro tiers, new apps in the same ecosystem), but I definitely feel like I’m swimming against the “everything is a subscription now” current.

What I’d love feedback on

This is the part where I’d really appreciate honest opinions from people here.

If you were in my shoes:

1) How would you price this kind of thing?

Pure one-time per app?
A bundle unlock (e.g. “FanStudio Offline Pack”)?
One-time per app + a separate “pro pack” for power users?
Something else I’m not thinking of?

I’d like to avoid classic monthly subs if possible, but I also don’t want to trap myself in “sell once, support forever”.

2) Would you lead with “offline AI” as the main pitch?

Right now I’m doing:

Main hook = concrete pain
- “Your iPhone storage is full”
- “Your documents are a mess”
Offline / privacy is the trust angle on top

Curious what you’d do:

Would you put “offline AI” front and center in the marketing?
Or keep it as a strong secondary point?

3) Any good examples of similar products / playbooks?

If you know indie products that:

Are utility-style
Care about privacy
Don’t rely heavily on subscriptions

…I’d love to study how they’re doing it.

Where I hope this goes

A year from now, I’d love FanStudio to be:

A small, sustainable portfolio of 3–5 offline AI utilities
Each one solving a very specific “annoyance”
All sharing the same values: on-device, transparent pricing, minimal friction

I’m based in Asia, building mostly for a global audience, and I’m basically trying to answer:

“Can a solo dev make a decent living from small, opinionated offline tools?”

Right now the honest answer is:
“Maybe? I hope so?” 😅

If you’ve tried something similar (or think I’m over-optimizing for principles and under-optimizing for survival), I’d really love to hear your take.

Also happy to share more concrete details (Core ML setup, model choices, any early numbers, mistakes) if that’s interesting.

Thanks for reading 🙏