On-device AI · Zero network · iOS 26

Your ideas, yours alone.

IdeaNook uses on-device AI to capture, categorize, and summarize everything you think — your data never leaves your phone. No cloud. No servers.

Join the TestFlight beta

iOS 26.0+ · Android & HarmonyOS in progress

Your ideas, yours alone.

Fully on-device AI

Gemma 3n E2B multimodal model runs locally on your device — zero network requests.

Privacy by design

Data lives only in the app sandbox. No network permission, ever uploaded.

Chat-style capture

Just text yourself — drop in text, photos, or voice with one tap.

Chat-style capture

As simple as texting yourself

When inspiration strikes, just send it. Text, photos, and voice all flow into one idea stream — zero learning curve.

Jot it down
Text

Jot it down

Send a thought the moment it arrives — natural as a chat.

Photo

Snap what you see

From camera or library in one second, with full-screen zoom.

Can't type? Just say it
Voice

Can't type? Just say it

Hold to record, release to send. AI organizes it for you.

On-device auto-categorization

Send and forget — AI files it for you

After each text message, on-device AI decides its category instantly — no waiting, no data leaving your phone.

1

Send

Write down any thought

2

Sorting

A “⏳ sorting” chip appears

3

Tagged

Turns into a real category tag

4

Browse

See all ideas of one kind

10 smart categories, all inferred locally

💼Work
📖Study
🏠Life
Ideas
💻Tech
❤️Health
$Finance
👥Social
✈️Travel
Other
AI summaries

What did you record this week? AI turns it into a digest

Reviews that actually matter. Every digest is generated on-device — the content never leaves your phone.

Automatic weekly digest

Automatic weekly digest

A Sunday-evening notification reminds you; reopen the app to auto-generate last week's review. Daily digests optional.

Trigger anytime

Trigger anytime

Today, this week, or a custom range — generate a summary whenever you want.

Summarize from here to now

Summarize from here to now

Long-press any message to build a tailored review of any time span.

Liquid Glass input

An input bar redesigned for iOS 26

Built on iOS 26's native Liquid Glass — three elements flow and breathe as one. Switch text ⇄ voice in a tap; the + menu morphs into the send button intelligently.

  • Native .glassEffect() fluid material
  • Text/voice modes switch with SF Symbol replace animation
  • Hold to talk enters a clear red recording state
Privacy First

Your thoughts stay on your phone

Privacy isn't a feature — it's the baseline. IdeaNook makes data leaks architecturally impossible.

Zero network requests

The app requests no network permission, so it simply cannot upload data.

Data never leaves

Everything lives in the app sandbox; even system backups stay under your control.

AI runs locally

Gemma 3n E2B runs on your iPhone — no cloud API is ever called.

Minimal permissions

Photos write-only, microphone & camera on demand — used then released.

Your thoughts stay on your phone
Under the Hood

Built for the curious

An all-Apple native stack with on-device multimodal inference. We keep the complexity so you get the smoothness.

01

Gemma 3n E2B multimodal

Google's on-device multimodal LLM — text + vision + audio.

02

LiteRT-LM engine

C++ runtime wrapped in a Swift package, accelerated by Metal GPU.

03

SwiftUI + SwiftData

An all-Apple native stack using the latest iOS 26 APIs.

04

iOS 26 Liquid Glass

Native .glassEffect() / GlassEffectContainer APIs.

05

Lazy thumbnail pipeline

CGImageSource generates thumbnails without decoding originals — buttery lists.

06

Local SwiftData sandbox

All data stays in the app sandbox, with optional encrypted iCloud backup.

Start capturing every idea

Join the TestFlight today and be first to try a fully on-device, truly private idea journal.

Join the TestFlight beta