Back to database

Andrew Wilkinson, Tiny's AI use case

Cofounder at Tiny

Built a set of personal AI apps for everyday life and work, including a relationship/personality analyzer, a custom email client, a meeting-pattern detector, and a wardrobe system that texts outfit ideas each morning.

The problem

What was broken before AI

A lot of personal friction is too specific for off-the-shelf software. Andrew was dealing with hundreds of emails a day, school messages that mattered only occasionally, meetings with subtle interpersonal dynamics, uncertainty about what to wear, and relationship patterns that were hard to see clearly in the moment. None of those problems justified a big software project, but each one created small drag in daily life.

What changed

What the use case made possible

Claude Code made it possible for Andrew to build small tools around those exact problems. His email system filters and triages messages, then offers structured reply options. His parenting workflow pulls important school details out of email and turns them into reminders. His meeting agent looks for red flags in transcripts. His wardrobe system uses a spreadsheet of photographed clothes, the weather, image generation, and text messaging to suggest outfits. The result is a set of highly personal apps that would probably never exist as normal SaaS products.

Why this matters

Why this use case is worth studying

Andrew’s use case shows what happens when software becomes cheap enough to make for an audience of one. Instead of waiting for a company to build the perfect tool, he can turn a recurring annoyance into a small working app. Some of these tools are practical, some are strange, and some are deeply personal. Together they point to a world where more software is custom-built around the way one person actually lives.

Use this when

When this pattern applies

Use this pattern when a recurring annoyance in your life is too specific for normal software, but clear enough that a small custom tool could help. It works especially well when the input is already available somewhere — email, calendar, transcripts, photos, weather, spreadsheets, or messages — and the output is something simple you would actually use.

Exponential Builder analysis

01

Build for an audience of one first.

Andrew’s apps work because they start with friction that is too specific for normal SaaS: school emails, meeting dynamics, outfit choices, inbox triage. AI coding tools make those edge-case workflows worth building because the bar is usefulness, not market size.

02

The input matters more than the interface.

Each tool becomes useful only after it connects to the messy source of the problem: Gmail, transcripts, weather, photos, spreadsheets, or text messages. Personal AI apps usually fail when they start with a chatbot and succeed when they start with the real data already sitting in someone’s day.

03

Keep judgment-sensitive tools on a leash.

Meeting analysis, relationship insight, and email replies can help surface patterns, but they should stay in a review-and-reflection role. The more a tool touches trust, tone, or another person, the more it needs thresholds, approval steps, and humility built in.

Who this is for

Best fit

Founders and operators with repetitive personal workflows

People who want tiny tools instead of full SaaS products

Builders experimenting with Claude Code or Replit

Busy parents managing reminders, email, and schedules

People with recurring decisions they want to simplify

Anyone curious about building personal software for an audience of one

What to avoid

Mistakes and warnings

Where this pattern can go wrong if you copy it too literally.

Do not connect private accounts or personal data without thinking through privacy and permissions.

Avoid letting AI make relationship, work, or health decisions for you.

Keep approval steps around anything that sends messages or affects another person.

Do not overbuild the first version; most personal tools only need to solve one small job.

Watch for confident analysis in emotionally charged situations; use it as a signal, not a verdict.

Public workflow preview

The shape of the workflow

A high-level look at how the use case works, with the reusable pattern made clear.

01

Notice the recurring friction

Andrew starts with a small problem that keeps showing up in daily life, like email overload, school reminders, relationship patterns, or outfit decisions.

02

Describe the job in plain English

Instead of writing a full product spec, he explains what he wants the app to do and how it should behave.

03

Build a tiny first version

Claude Code helps turn the idea into a working web app, agent, or automation that can be tested quickly.

04

Connect the real-world inputs

The tools pull from practical sources like Gmail, meeting transcripts, Google Sheets, weather, photos, or text messages.

05

Use the tool in daily life

The value shows up when the app becomes part of an existing routine, such as checking email, preparing for meetings, or getting dressed in the morning.

Copy the pattern

The reusable idea

Pattern in one sentence

Turn small recurring annoyances into tiny AI-powered tools that fit directly into the routines where those annoyances already happen.

Reusable idea

Andrew’s use case is a reminder to look for the tiny problems that are too personal for normal software. The best place to start is often a repeated irritation: the email you sort every day, the reminder you keep missing, the decision that takes more energy than it should. If the input is clear and the output is useful, AI can help turn that annoyance into a small tool built around your life instead of everyone else’s.

Steal this workflow

Pick one annoyance that repeats at least weekly and write a one-page mini-spec:

Friction: What keeps happening that drains time, attention, or energy?

Routine moment: When exactly would you use the tool?

Inputs: What would the app need to see? Examples: emails, calendar events, transcripts, photos, weather, a spreadsheet.

Output: What should it hand back? Examples: reminder, ranked list, draft reply, warning, summary, recommendation.

First version: What is the smallest version that would be useful tomorrow?

Data limit: Connect one or two sources only.

Review rule: Where must a human approve before anything is sent, shared, or acted on?

Keep-or-kill test: After one week, did it save time, improve clarity, or reduce a recurring decision? If not, simplify or delete it.

Suggested prompt

“I want to build a small personal AI app for this recurring problem: [describe the annoyance]. It happens during this routine: [when/how it shows up]. The app should use these inputs: [emails/transcripts/calendar/photos/weather/spreadsheet/etc.]. It should produce this output: [reminder, ranked list, draft reply, summary, warning, recommendation]. Build the simplest working version first, with only the minimum data connections needed. Add a human review step anywhere the output could affect another person, send a message, or influence an important judgment.”

Field notes

Get new AI use cases in your inbox

A short weekly note on how real people are using AI to save time, make money, build tools, and run their lives.

No spam. Just useful AI use cases.

Related use cases

Keep exploring nearby systems.

Browse all