Shelf.app

This week I built four little apps inside Shelf — the AI desktop-app generator I'm shipping — for problems I wasn't planning to solve. None of them could be a SaaS. Each took 20–40 seconds to generate. Combined, they save me roughly 40 minutes a week.

Posting them here partly because IH readers will recognise themselves in one or two, partly because "I dogfood my own product" is easy to claim and harder to show. Here's the showing.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
## 1. Standup-to-Self

Prompt:

> "End-of-day check-in. One textarea for 'what I worked on today' and one for 'where I'm picking up tomorrow'. Save with date. On open, show yesterday's 'tomorrow' note at the top in big text."

Why I needed it. I was losing about 10 minutes every morning re-orienting on what I'd been doing the day before. A Notes app didn't solve it because I had to remember to open Notes. With Shelf already in my dock, the note is the first thing I see.

Why no SaaS builds this. A SaaS that solves this has to also be a journal, a planner, a habit tracker, a "system." Mine is one textarea and a date column. It will never be a startup. It's been perfect since I built it.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

## 2. Tweet Draft Queue

Prompt:

> "Tweet ideas list. Add text, optional tag (ship / thinking / snark / data). Mark as posted with link. Show only unposted by default, sorted oldest first."

Why I needed it. I have tweet ideas all day and lose 90% of them by evening when I'd actually post. Notion would technically work; opening Notion takes me four seconds and a context-switch. Shelf is already open. That friction delta is the difference between actually posting and not.

Why no SaaS builds this. Hypefury, Typefully, Tweet Hunter — they all do this. Each costs $20–$50/month and includes 40 features I don't use. The tax for needing one feature is the other 39.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

## 3. MRR What-If Calculator

Prompt:

> "Pricing calculator. Sliders for visitors per day, free-to-paid conversion %, monthly churn %, average revenue per user. Live MRR projection for month 3, month 12, and steady-state. Toggle: 'what if I doubled visitors'."

Why I needed it. I keep rebuilding the same spreadsheet to sanity-check pricing decisions. The spreadsheet keeps adding parameters and getting harder to compare to last week's version. This is the spreadsheet, except it's an app, it's persistent, and the sliders feel right.

Why no SaaS builds this. ProfitWell-class tools exist, attached to actual revenue data. Mine is for hypothetical scenarios — playing with numbers before committing to anything. Different job, no market, perfect Shelf use case.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

## 4. Founder Mood Check

Prompt:

> "Once-a-day check-in. Question: 'how's it going?' (1–10 slider) + one-line text 'why'. Save with date. Show the last 30 days as a sparkline."

Why I needed it. Building solo for a year does interesting things to your sense of how the project is going. I wanted a sparkline of me, not the metrics. Every week I look at it I can tell whether last Tuesday's 3-out-of-10 was the start of a real dip or just a bad Tuesday.

Why no SaaS builds this. There are 4,000 mood-tracker apps in the App Store. None are built for a founder context, and all of them want my phone notifications and a freemium upgrade flow. Mine asks one question and draws the line.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

## What this actually tells you about Shelf — and about your own product

A SaaS founder usually has to choose between the thing you're building for customers and the thing you'd build for yourself if you had infinite time. Shelf collapses that choice for me. The customer-facing product and my own quality-of-life tools come out of the same prompt box.

There's a wider read for IH readers thinking about their own products: when your product is cheap enough for users to deploy on themselves, the "do they actually love this?" question gets a brutally honest answer. They start using it for things you never built it for. You feel it the day they do.

Which is, increasingly, the only version of dogfooding I trust.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

## Try one

- Download Shelf: [getmyshelf.app](https://getmyshelf.app) — free, desktop, BYOK or use the hosted free tier.

- Pick the prompt above that looks most like a problem you'd actually use this week. Paste it in. The point isn't to use my four apps. The point is to feel how short the loop is between "I wish I had X" and "oh, I have X."

- Post yours back here. I'd genuinely like to see what other IH builders make when they don't have to justify it to a roadmap.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Shelf is a free desktop app. Data stays on your machine. The point isn't to replace big software — it's to fill the gap big software was never going to fill.

A nurse I know finishes her ward shift at 7am. The 15 minutes before that are the most important of her night — handover. She tells the day shift what happened overnight, what to watch, what the on-call doctor said, which patient's family is coming in upset, which IV pump is about to alarm.

The hospital has a multi-million-pound EHR. None of it captures any of that.

So she does what every nurse on every ward does: she scribbles on a piece of paper. Her own format. Her own shorthand. "Bay 3 — K+ recheck 8am — bloods chased — daughter v anxious." Then she hands the paper to the day shift, who bins it by lunch.

Last week she opened Shelf and typed:

> "Shift handover notes. Add patients by bay number with name, key issues, tasks to do, things to watch overnight. Tick off tasks. One button to clear everything at end of shift."

About 30 seconds later Boom! she had it. Running on her laptop. Data storage under the hood. No login. No cloud. No PHI on anyone's server because there is no server.

## Why no SaaS was ever going to build this

Indie hackers know this math cold. To build a "handover notes app for nurses" as a real product you need:

- HIPAA / GDPR compliance, audit logging, encryption at rest and in transit

- EHR integration (Epic, Cerner — either will eat you alive for a year)

- A procurement motion into hospital trusts (18 months, minimum)

- Liability insurance to sign the contract

- A team large enough that the unit economics demand £40k/seat/year

So the only software that exists for nurses is the multi-million-pound EHR — which doesn't have a "things to tell the next nurse" field, because that field has zero billable value.

The actual need — I want private notes, on my own device, in my own format, that wipe at end of shift — has been sitting in the long tail since 1990. Worth nobody's time to build. Worth one nurse's time to describe in a sentence.

## What changed

The cost of building one personal tool used to be "learn to code, spin up a server, host it, maintain it forever." Now it's one sentence in plain English.

That's what Shelf is. A free desktop app that takes that sentence, generates a working HTML app, and runs it locally with persistent storage. No account. No cloud. The SQLite file lives on her laptop. If the ward wifi drops (it does), the app keeps working. If she switches hospitals next year, the app and every note she's ever taken come with her.

This is the long tail of software finally getting a tail. The market for a generic nurse-handover app was always 1.5 million nurses × zero willingness to onboard to yet another tool. Multiply by zero, you get nothing — which is exactly what's been built. The market for each nurse's own tool is one person, and the cost per build is now lower than the cost of complaining about the EHR.

The same mechanism that built her handover app also builds:

- A patient-bay round timer for a hospice nurse

- A daily P&L tracker for a corner shop

- A revision quiz scoped to one chapter of one textbook

- A pit-stop trainer for a sim racer

None of these are startups. All of them are 30 seconds of typing.

## Try it

- Download Shelf: [getmyshelf.app](https://getmyshelf.app)

- The exact prompt: "Shift handover notes. Add patients by bay number with name, key issues, tasks to do, things to watch overnight. Tick off tasks. One button to clear everything at end of shift."

- Or skip the build: import [shift-handover-notes-add-pa.shelf](https://raw.githubusercontent.com/soyu-apps/getmyshelf.app/refs/heads/main/samples/shift-handover-notes-add-pa.shelf) — drag it into the sidebar, her exact app appears on your machine.

Show it to a nurse you know. See how long it takes them to ask if they can keep it.

The answer is yes. That's the whole point.

---

Shelf is a free desktop app. Apps you make in it stay on your machine forever. The point isn't to compete with big software — it's to fill the gap big software was never going to fill.

Every "F1 lights reaction tester" on the internet is broken. Not buggy — just wrong about what F1 drivers actually do.

You know the drill. Five red lights come on. They go out. You click as fast as you can. Some site tells you your reaction was 230ms and to try again.

That isn't what F1 drivers do.

A real driver sits at the grid with the clutch already pulled in. When the lights go out, they release the clutch. The reaction isn't a click — it's a release.

My 8-year-old noticed this. He watches every race. He kept asking why none of the online ones felt right. So he opened Shelf and typed:

> "F1 start lights reaction test. Hold spacebar before lights, release when they go out. Show reaction time."

12 seconds later he had it. Working. On his laptop. Forever.

## Why no website builds this

It's not hard to build. It's hard to justify building.

A solo dev wanting to ship a "reaction time" tool picks the click version because:

- 95% of visitors don't know about the clutch thing

- Adding a hold-to-release mode adds settings, edge cases, mobile fallback

- Every extra feature is one more thing to maintain for the next 10 years

- There's no money in F1 reaction testers anyway

So you get the same generic version everywhere. The 5% with a specific need — kids who watch races, sim racers, motorsport fans — get nothing.

This is the customisation tax. Every piece of software you use is a compromise between what you want and what was profitable to ship to a million people who mostly want the same thing.

## What changed

The economics changed. The cost of a one-person tool used to be "learn to code, spin up a server, host it, maintain it." Now the cost is one sentence in plain English.

That's what Shelf does. You describe a tool — even a weird, hyper-specific one only you and your kid care about — and it builds it as a real app on your machine. SQLite-backed. Offline. Forever. No account, no hosting, no JS framework to learn.

It scales the other way too. The same prompt mechanism that made my kid's reaction tester also makes:

- A patient handover form for a nurse

- A daily P&L tracker for a small shop

- A revision quiz scoped to one chapter of one textbook

- A 5-minute classroom timer with the exact sound that teacher likes

Every one of these is a real tool somebody needs. None of them are big enough for a startup to build. All of them are 30 seconds of typing in Shelf.

## Try the F1 reaction tester

I exported my kid's app as a .shelf file. Download it, install Shelf, drag the file into the sidebar — that's the import. His exact app, including the spacebar-release mechanic, runs on your machine.

- Download Shelf: [shelf.app](https://shelf.app)

- Download the F1 reaction tester: [f1-reacting-lights.shelf](https://raw.githubusercontent.com/soyu-apps/getmyshelf.app/refs/heads/main/samples/f1-reacting-lights.shelf) (attach when posting)

- Or build your own: open Shelf and type "F1 start lights reaction test, hold spacebar before lights, release when they go out, show reaction time"

Show it to your kid. See if they can beat his time.

---

Shelf is a free desktop app. Data stays on your machine. The point isn't to replace big software — it's to fill the gap big software was never going to fill.

A teacher I know spent £4/month on a classroom timer app she used twice a day. A nurse kept patient handover notes on the back of a printout because no app fit her shift. A bakery owner underquoted a wedding cake by £80 because she did the maths in her head.

Every existing tool does 90% more than they need, charges monthly, or demands learning a new interface.

So I built Shelf.

You type what you want — "countdown timer with a label, rings when done, remembers my last 5 labels" — and it appears on your screen in under 10 seconds. Lives on your computer. Works offline. Saves your data forever. No account. No cloud. No subscription.

The teacher built a random student picker in 20 seconds and has used it every day for a month. The nurse made a shift handover log that survives between shifts. The bakery owner built a quote calculator and saved the recipe for "Lucy's wedding cake" — every ingredient cost still there six weeks later.

None of them know how to code.

The weird part is what happens after. People build one small thing — a habit tracker, a book journal, a grocery price log — and then they build another. And another. They text screenshots to friends. They ask "can it do X?" and the answer is always yes, because the next app is also 10 seconds away.

Here's the deal. The app is free on your computer — generate as many tools as you want, keep them forever, your data never leaves your machine. That part never costs anything.

The paid plan only matters if you want your apps and data to follow you across devices — laptop to phone, work to home. For that, your stuff needs somewhere to live. That's what the lifetime plan covers: £79 once, 1GB of cloud storage, unlimited apps, yours forever. No monthly fee. No renewal.

Founder offer — ends May 31: lifetime drops to £29 (60%+ off). After that, it's £79 forever. No second chance.

If you have ever wanted a tool that does exactly the way you want it — a tracker, a timer, a log, a calculator, a checklist — you can have it tonight. Free version costs nothing. Founder price won't come back.

Try it before your friend sends you a screenshot of theirs.

https://getmyshelf.app