Burrito’s Labs Indie Tech Outmost Studio
Personal Instruments
Active development
Drop app icon with a dark teardrop form and a small blue vertical mark on a white rounded square.

Drop

A local-first iPhone inbox that captures text, voice transcripts, and readable images, then turns them into saved notes, summaries, checklists, reminder candidates, and retrievable ideas.

Category
Personal Instruments
Platform
iPhone · iOS 17+
Status
Active development

What if capture did not require an immediate filing decision?

Typed, pasted, spoken, and shared input enters one local inbox. Deterministic transformations, search, related items, reminder review, export, and deletion make each fragment useful and controllable.

User value

Capture first. Decide what it becomes later.

Give unfinished thoughts, useful snippets, voice transcripts, and readable images one place to land before their purpose is clear.

01

Catch the fragment

Type, paste, dictate, or share text and readable images without choosing a permanent structure first.

02

Shape it progressively

Keep the original while saving a separate Clean Note, Summary, Checklist, or Reminder Candidate.

03

Find and control it

Search the source and saved outputs, surface related Drops, then archive, export, restore, or remove the material locally.

Drop iPhone onboarding screen introducing quick capture, useful transformations, related ideas, local-first storage, and a Get Started button.View full size
Current introductionCurrent Debug simulator capture presenting the product's four core promises before any permission request.

Features

From raw input to something useful.

The current app connects capture, local persistence, transformation, retrieval, action, and data control.

  1. 01

    Multi-source capture

    Save typed or pasted text, record and transcribe a voice note through Apple speech services, or receive shared text from another iOS app.

  2. 02

    Readable-image ingestion

    Recognize text from a shared image with Apple Vision and retain a local original and thumbnail only when usable text is found.

  3. 03

    Four persisted transformations

    Create Clean Note, Summary, Checklist, and Reminder Candidate outputs without replacing the captured source.

  4. 04

    Reviewed reminder handoff

    Confirm or correct supported date and time text before asking Drop to create an Apple Reminder.

  5. 05

    Search and Related Drops

    Match active items across original text, recognized text, and saved outputs, with experimental local semantic suggestions ranked below direct matches.

  6. 06

    Visible data controls

    Archive, restore, permanently delete, export, inspect local storage, maintain the semantic index, delete all data, or reset onboarding.

How it works

Catch it now. Structure it when useful.

One Drop can move from unfinished input to several useful readings without losing its source.

  1. 01

    Capture from the current context

    Use the Capture tab, clipboard, voice flow, or iOS Share Extension to preserve the material quickly.

  2. 02

    Store the original

    Create one local DropItem with its source and date. Readable images can also retain managed local preview files.

  3. 03

    Choose a useful output

    Clean the text, summarize it, extract a checklist, or create a reminder candidate through the default deterministic backend.

  4. 04

    Review before handoff

    Inspect saved output and confirm or correct any supported reminder date before using Apple Reminders.

  5. 05

    Retrieve it later

    Search source and output text or use optional local embeddings to see conceptually related active Drops.

  6. 06

    Control the lifecycle

    Archive, restore, export, permanently delete, or remove all local Drop data through explicit confirmed actions.

Current interface

A focused inbox for active Drops.

The current Inbox keeps search, source, capture date, and the three primary product areas visible without requiring folders at capture time.

Drop iPhone Inbox with a search field and two non-personal seeded items labeled Shared Text and Typed above Inbox, Capture, and Settings tabs.View full size
Current Debug simulator capture using fixed non-personal seed data.

Seeded active Inbox

Direct search sits above two deterministic, non-personal development items with their capture source and date.

Product concept

Incomplete input is a valid starting point.

A person should be able to preserve a passing fragment before deciding whether it belongs in a note, task, reminder, or longer system.

01

Capture before classification

Speed matters when the thought is passing, so Drop begins with the input instead of a required destination or folder.

02

Outputs do not erase sources

Each transformation remains a separate saved interpretation, allowing the original fragment to support another use later.

03

Retrieval completes the loop

Search, related items, archive, export, and deletion determine whether quick capture becomes dependable personal memory.

Architecture

One local model connects the app and Share Extension.

The SwiftUI app and UIKit extension share an App Group persistence boundary, while focused services own capture, transformation, retrieval, system handoff, export, and cleanup.

  1. 01App or Share Extension input
  2. 02Validation, transcription, or OCR
  3. 03SwiftData item and local media
  4. 04Persisted transformation output
  5. 05Keyword and optional semantic retrieval
  6. 06Reminder, export, archive, or deletion

Shared local persistence

The app and extension write through the same App Group boundary without a project-owned server or account system.

Media stays outside the database

SwiftData keeps references and metadata while managed files hold readable-image originals and thumbnails for explicit cleanup.

Deterministic behavior remains available

Heuristic transformations and direct keyword matching are the production defaults, with semantic-only matches ranked below factual text matches.

Optional model work stays gated

Apple Foundation Models support is conditional, malformed output is rejected, heuristics remain the resolver default, and real-device behavior is not yet validated.

Technology

Native, local, and service-bounded.

Application

  • Swift
  • SwiftUI
  • UIKit Share Extension
  • iOS 17+

Capture

  • AVFoundation
  • Apple Speech
  • Vision OCR
  • ImageIO thumbnails

Local data

  • SwiftData
  • App Group container
  • Managed media files
  • UserDefaults

Retrieval and action

  • In-memory keyword search
  • NaturalLanguage embeddings
  • EventKit reminders
  • System share sheet

Current boundary

  • No project backend
  • No account or cloud sync
  • No analytics
  • No remote AI API
  • Optional Foundation Models gated

Development

The capture-to-retrieval loop is functional.

The native app, Share Extension, capture paths, transformations, search, related items, reminder review, archive, export, media cleanup, and local data controls are implemented and broadly test-backed. Signing, final hardware validation, system integration checks, accessibility, appearance, and optional local-model hardening remain incomplete.

Stage
Active development
Unit tests
278 passed
Platform
iPhone · iOS 17+
Public release
Not confirmed

Implemented

  • Onboarding, Inbox, Capture, Item Detail, Archive, Search, Related Drops, and Settings surfaces
  • Typed, pasted, voice-transcript, shared-text, and readable-image capture paths
  • SwiftData and App Group persistence with local originals, thumbnails, previews, and lifecycle cleanup
  • Clean Note, Summary, Checklist, and Reminder Candidate outputs persisted separately from the source
  • Keyword search, experimental local embeddings, direct-match precedence, and semantic index controls
  • Reminder review and EventKit handoff, local storage summary, text and JSON export, restore, permanent deletion, and delete-all

Before release

Resolve signing and provisioning, run the signed-device smoke matrix, repeat readable-image sharing, export, reminder, VoiceOver, Dynamic Type, light mode, and dark mode checks on hardware, then harden validation, fallback, user-facing status, performance, privacy copy, and stability for any optional Foundation Models path.

Drop lab notes

The project, over time.

Notes about the thinking, architecture, development, and decisions behind this work.

  1. Drop
    3 Jun 2026 · Draft

    Safer Reminders and an Optional Local AI Boundary

    Explicit date review and a model-agnostic backend boundary protected dependable behavior while optional local intelligence remained gated.
  2. Drop
    2 Jun 2026 · Draft

    Making Local Storage Visible and Controllable

    Managed image files, archive recovery, export, storage summaries, and deletion controls made local-first behavior inspectable.
  3. Drop
    20 May 2026 · Draft

    From Stored Fragments to Resurfacing Ideas

    Direct search, local embeddings, and Related Drops began bringing useful material back without obscuring why it matched.
  4. Drop
    30 Mar 2026 · Draft

    Establishing the Capture Loop

    A common local DropItem let typed, pasted, spoken, and shared input enter one capture-to-inbox loop.