Join the waitlistDownload

How to brief an AI agent to build version 1 of your iPhone app

What goes into a brief for a coding agent, a full example for a guitar fretboard trainer, and how to steer the agent after version 1.

In short

A good brief for an AI agent fits in ten lines. What you're building and with what, who it's for, how it differs from two or three named leaders, the five screens of version 1, what version 1 leaves out, how the app makes money and what to check before you start. All 50 briefs in the Appto book 50 iPhone app niches follow this order (October 13, 2026).

Why version 1 has five screens

Version 1 does one job. Five screens are enough for someone to open the app, do the main thing and see the purchase screen. Everything else waits for version 2.

The narrower the brief, the less the agent has to guess. It doesn't spend time on accounts, feeds and sync nobody asked for, and it gets to a build you can open in the simulator sooner. That's why all 50 briefs in the Appto book start with a five-screen version 1 and spell out what it leaves out.

Version 1 does one job: five screensSketches of the five version 1 screens of a guitar fretboard trainer from the plan for niche 01 in the book 50 iPhone app niches: 1. fretboard with notes, 2. scale and chord picker, 3. a timed drill with a question like find C# on the 5th string, 4. progress by fretboard zone, 5. settings and tunings. Next to them, what waits for version 2: songs, tabs and social features from the brief, accounts, feeds and sync from the article. Book edition of October 13, 2026.Version 1 does one job: five screensA guitar fretboard trainer, niche 01 in the book 50 iPhone app niches1. Fretboard withnotes2. Scale andchord pickerC#3. Timed drill4. Progress byfretboard zone5. Settings andtuningsWaits for version 2SongsTabsSocial featuresAccountsFeedsSync
Five screens are enough to open the app, do the main thing and reach the purchase; everything else waits for version 2. Appto sketches from the book's plan, niche 01, October 13, 2026 edition, not screens of an existing app.

The seven parts of a brief

Every brief in the book follows the same order. Each part settles a question the agent would otherwise answer on its own, probably not the way you would.

  1. What you're building and with what. One sentence. A native iPhone app in SwiftUI and the core idea in a few words.
  2. Who it's for. One person with one job. A guitarist learning notes on the neck. A collector going through a box of cards.
  3. How it differs from named leaders. Two or three apps by name and three or four concrete differences. The agent gets a target, and the app doesn't come out as a copy of the leader.
  4. The five screens of version 1. Screen names in the order a person sees them.
  5. What version 1 leaves out. An explicit list. Without it the agent adds extras with good intentions.
  6. Money. What's free, what's paid, one purchase or a subscription, ads or none.
  7. What to check before you start. Whether the name is free on the App Store, where the data comes from, whether anyone else holds rights to the content.

Sometimes there's an eighth line about content. For a party game it's 600 to 900 cards across 8 to 12 decks that you write yourself, with no photos or camera. For a weekly meal planner it's no recipe database, people enter their own dishes (Appto book, October 13, 2026).

A full example. A guitar fretboard trainer

Here is the brief for niche 01 from the book, line by line.

PartLine of the brief from the book
What and on whatBuild a native iPhone app in SwiftUI: an interactive guitar fretboard trainer.
Who it's forGuitarists learning notes, scales and chords on the neck.
How it differs from Solo, Ultimate Guitar and FretonomyTimed drills like "find C# on the 5th string", a score by fretboard zone, any tuning, fully offline, no account.
Version 1, five screensFretboard, scale and chord picker, timed drill, progress by zone, settings with tunings.
Not in version 1Songs, tabs, social features.
MoneyOne purchase or a yearly Pro plan, no ads.
Before you startCheck that the app name is free on the App Store.

Appto book 50 iPhone app niches, niche 01, October 13, 2026 edition.

Seven lines, and the agent has little room to guess. It knows who the app is for, what it will be compared with, which screens to draw and what to leave alone. The money and the market in this niche are covered in the guitar fretboard trainer write-up.

A whole brief for the agent: seven linesA brief for an AI agent from the book 50 iPhone app niches, niche 01, a guitar fretboard trainer, October 13, 2026 edition. Seven lines and what the agent learns from each. 1. Build a native iPhone app in SwiftUI: an interactive guitar fretboard trainer, what to build and with what. 2. Guitarists learning notes, scales and chords on the neck, who the app is for. 3. Differs from Solo, Ultimate Guitar and Fretonomy: timed drills like find C# on the 5th string, a score by fretboard zone, any tuning, fully offline, no account, what it will be compared with. 4. Version 1, five screens: fretboard, scale and chord picker, timed drill, progress by zone, settings with tunings, which screens to draw. 5. Not in version 1: songs, tabs, social features, what to leave alone. 6. Money: one purchase or a yearly Pro plan, no ads, how the app makes money. 7. Before you start: check that the app name is free on the App Store, what to check first. Sometimes an eighth line on content, for a party game 600 to 900 cards in 8 to 12 decks you write yourself.A whole brief for the agent: seven linesNiche 01 in the book 50 iPhone app niches, October 13, 2026 editionThe agent knowsBrief, guitar fretboard trainer1Build a native iPhone app in SwiftUI: an interactive guitar fretboard trainer.what to build and with what2Who it's for: guitarists learning notes, scales and chords on the neck.who the app is for3What makes it different from Solo, Ultimate Guitar and Fretonomy: timeddrills like "find C# on the 5th string", a score by fretboard zone, any tuning,fully offline, no account.what it will be compared with4Version 1, five screens: fretboard, scale and chord picker, timed drill,progress by zone, settings with tunings.which screens to draw5Not in version 1: songs, tabs, social features.what to leave alone6Money: one purchase or a yearly Pro plan, no ads.how the app makes money7Before you start: check that the app name is free on the App Store.what to check firstSometimes an eighth line on content: for a party game, 600 to 900 cards in 8 to 12 decks you write yourself
Seven lines, and the agent has little room to guess: it knows who the app is for, what it will be compared with, which screens to draw and what to leave alone. Appto book 50 iPhone app niches, niche 01, October 13, 2026 edition.

What to leave out of a brief

  • "Make it like the leader." Name the differences in words. One more lookalike with a new icon gets lost among the others.
  • Other companies' trademarks in the name. Pokémon and Magic don't belong in the name or subtitle of a card scanner, even if some competitors put them there.
  • Other people's content. Recipes on other sites are copyrighted. Bible translations have to be public domain, such as KJV and WEB; NIV and ESV are licensed (Appto book).
  • Everything at once. Move accounts, feeds, chat and sync to version 2.
  • Promises the app won't keep. In astrology, interpretations are framed as something to reflect on, never as predictions.
  • Passwords and keys. A brief doesn't need them.

What to check before you start

The last line of the brief is the shortest, and the most expensive to skip. Examples from the book (October 13, 2026 edition).

  • The name. Verseful and VerseMark are taken in Bible verse memorization, Vaulty is taken for a photo vault, Sweepshot is taken twice. The word charades is overcrowded in search.
  • The data. For Magic cards there's the open Scryfall database, which Appto checked on September 24, 2026. Some Open Library editions have no ISBN, so adding a book by hand has to be a full path.
  • Calculation accuracy. An astrology app's planet positions get checked against a published ephemeris before any screens are built.
  • Season. Yearly reading goals are set in January, so a reading tracker's launch should be planned around that.

How to steer the agent after version 1

Once the first build is ready, the editing begins. This is where a human eye matters.

  • Look at every screen. The agent checks the screens in the simulator, and a human eye catches what a check can't. In Appto a step gets its checkmark only after an actual build and check.
  • One change per screen. Be specific about where. "On the drill screen the timer is too small, make it bigger and move it to the top." In Appto you can pin a note right on the screen mockup.
  • New ideas go on the version 2 list. The App Store and the first purchases come first.
How to steer the agent after version 1The editing loop after the first build. The agent builds the app and checks the screens in the simulator, and a step gets its checkmark only from a fact. You look at every screen, since a human eye catches what a check can't. You give one change per screen, specific about where, and in Appto you can pin a note right on the screen mockup, for example: on the drill screen the timer is too small, make it bigger and move it to the top. Then another build and check. New ideas go on the version 2 list: the App Store and the first purchases come first.How to steer the agent after version 1Once the first build is ready, the editing beginsAgentBuild and checkthe agent checks the screens inthe simulator, a checkmark onlyfrom a factYouLook at every screena human eye catches what acheck can'tYouOne change per screenspecific about where; in Appto, anote right on the mockupagainA change, for example: "On the drill screen the timer is too small, make itbigger and move it to the top."A new idea goes on the version 2 listthe App Store and the first purchases come first
The agent builds and checks, you look at every screen and give one change at a time with its place; new ideas wait for version 2. How Appto works as of October 13, 2026.

The whole path from brief to App Store is in how to build an iOS app with AI, and the release itself in how to publish an app on the App Store.

Where to get ready briefs

The book 50 iPhone app niches has 50 ready briefs for an agent, one per niche, next to the niche's revenue, leaders and their weak spots. Five niches with their briefs are free, including the guitar fretboard trainer and the party game on one phone.

A brief goes into Claude Code, Codex or Appto.

Questions

What language should the brief be in?

All 50 briefs in the book are in English. Agents understand other languages too; what matters more is that all seven parts are there.

Does the brief work with Codex?

The book suggests pasting it into Claude Code, Codex or Appto. The seven parts don't depend on the tool.

Why no accounts or social features in version 1?

Version 1 does one job on five screens to reach the App Store sooner. Accounts, feeds and sync wait for version 2.

How long should a brief be?

The briefs in the book run seven to nine lines. If yours is much longer, version 1 is probably doing more than one job.

50 niches checked for money

The Appto book: the money in each niche, the leaders and their weak spots, a version 1 plan and a ready task for the agent. 5 niches free.

Open the book