Wrap-up — pick a pattern and ship your first tool

The patterns side by side, a ship-day checklist, and ideas grounded in real people you already know.

12 min read·9 parts

Part 1 · The case

You just closed Lesson 6. You've seen five little apps, five different flows, five different security postures. And now you're staring at a blank chat with Lovable thinking the same thing everyone thinks: what should I actually build first?

That's what this lesson is for. No new project. Instead: a map of what you already know, a way to match it against ideas you have in your own life, and a short checklist so the first thing you ship is boring, safe, and useful — not clever and broken.


Part 2 · The road so far

Every project in this course started the same way — a real person with a real time-waster — and ended with an app that made that specific pain smaller. In order:

  • Project A · Activity Collector. Aina at an NGO wanted staff to log volunteer hours. The trick: anyone on a pre-uploaded list can submit; only the admin reads. Public write, private read.
  • Project B · Personal Timetable Generator. Sam the program admin burned an afternoon a semester copy-pasting rows from one master timetable into thirty personal ones. Upload the master once, get a filled blank template per lecturer — deterministic parsing where it can be, AI only for the messy day/time cells. Private tool, one owner.
  • Project C · Live Field Tally. Zara ran a food-relief drive where seven volunteers counted on paper and never agreed. Every "+1" tap on a phone became one row in the database; the coordinator's screen updated the instant any phone tapped. Many writers, live shared number.
  • Project D · Community Event Pre/Post Survey. Dr Farah wanted an honest "did this change anything?" answer after a one-day community talk. Same short survey before and after, one comparison dashboard — built either fully in Lovable Cloud or with Google Forms feeding a Lovable dashboard. Anonymous public write, private compare.
  • Project E · Study Tracker. Danish logged his study sessions before an exam and never wanted anyone else to see. The rule "only the owner sees their rows" lived inside the database, not the app screen. One private owner.

Five human problems. Five flows. And — this is the important bit — only five patterns. Almost every internal tool you'll ever build is a remix of one of these.


Part 3 · The five patterns, side by side

Read the table the same way each time: who is allowed to write, who is allowed to read, and where the rule that stops the wrong person lives. In every row the rule lives inside the database — that was the point of Lesson 1, and every project since has repeated it.

PROJECTWHO WRITESWHO READSWHERE THE RULE LIVESA · Activity Collectoranyone on allowlistadmin onlydatabase (email allowlist)B · Timetable Generatorthe one power usersame power userdatabase (owner-only)C · Live Field Tallyvolunteers with codecoordinator (live)database (event code)D · Event Pre/Post Surveyanyone at the eventevent owner onlydatabase or Google SheetE · Study Trackeryou (signed in)only youdatabase (owner-only)In every project, the guard is inside the database — never only in the app on the screen.
The five projects, one row each. The last column is the one that keeps you safe when things go sideways.
One idea, one sentence
The app on the screen is a polite receptionist. The database is the locked door behind it. If the receptionist gets tricked, the door still holds — only if you wrote the rule inside the database.

Part 4 · How to pick a pattern for your own idea

Before you open Lovable and start typing, answer these five short questions in your head. The first "yes" is your pattern.

  1. 1Does anyone in the world need to submit, but only a small team reads?

    Project A pattern. Public form, admin dashboard, allowlist if the submitters are known in advance. Examples: a feedback box, a volunteer sign-up sheet, a bug report form.

  2. 2Is it one power user cleaning up files over and over?

    Project B pattern. Sign-in, private workspace, AI does the boring first pass, human keeps the veto, export a clean file at the end. Examples: bulk-renaming photos, tidying a customer list, extracting fields from PDFs.

  3. 3Do lots of people need to watch the same number change live?

    Project C pattern. Short shared code, one row per event, coordinator screen that updates on its own. Examples: event headcount, class attendance, a giving thermometer.

  4. 4Do you need before-and-after answers from a group to see if something worked?

    Project D pattern. Same short survey asked twice, anonymous, dashboard compares group averages. Examples: a training workshop, a health talk, a school programme, a change-management rollout.

  5. 5Is the whole thing just for you (or one person at a time)?

    Project E pattern. Sign-in required, owner-only rule in the database, no sharing at all. Examples: a personal habit tracker, a private journal, your own reading list.

If none of the five fit cleanly, your idea is probably two tools glued together. Build the smaller half first.


Part 5 · A ship-day checklist

Boring but critical. Before you send the link to a single real person, walk this list out loud. If you can't answer a line, that's the next thing to fix.

  • Who signs in, and who doesn't? One sentence per audience. If you can't say it in one sentence, the app is trying to be two apps.
  • What is each audience allowed to see and do? And critically — where does that rule live? If the answer is "the button is hidden", that's not a rule. The rule lives in the database.
  • What happens when the answer is "no"? Not signed in → a polite sign-in screen, not a scary error. Not the owner → a "you don't have access to this" page. Not found → a "this doesn't exist" page. Never a blank white screen.
  • What happens when there's nothing to show? First-time users see empty tables. Write one friendly sentence per empty state ("No entries yet — add your first one above.").
  • Backups. Lovable Cloud takes daily backups automatically. Know that they exist. Know how many days back they go. That's usually enough.
  • Cost. Every project in this course fits comfortably in the free tier while you're learning. Before you invite 500 people, glance at the usage page. No surprises.
  • A stranger tries it on a phone. Not you, not your laptop, not your Wi-Fi. Hand your phone to someone who doesn't know the app and watch them use it silently for two minutes. Everything you were going to "explain later" becomes obvious in those two minutes.

Part 6 · Ideas for your first real tool

Not made-up business ideas — the small, boring, real ones you've already heard someone complain about. Each idea below names the pattern it fits so you can start from a lesson you've already read.

Do
For someone you know
Ustaz tracking hafazan. One private log per student, only the ustaz sees the whole class. Project D + a little admin view.

A kedai runcit's daily sales log. Owner signs in on the phone at night, types today's total. Simple graph at the end of the month. Project D.

A cikgu's attendance sheet. Students tap their name on a shared class link during the first two minutes. Cikgu sees who came, live. Project C.
Do
For a small team or event
Family reunion RSVP. One link, anyone can say "coming" and add their name. Only the auntie organising sees the list. Project A.

Wedding photo collector. Guests upload from their phones. Only the couple sees everything at once. Project A with a file locker attached.

Class quiz scoreboard. Students tap answers on their phones; teacher's screen shows the tally in real time. Project C.
The unglamorous rule
The best first project is one specific person who has already told you they're annoyed. Not a general audience. Not "the market". One person. Build the small thing that removes their specific annoyance. If they use it twice, you have a tool.

Part 7 · What we deliberately did not cover

Every one of these is a bigger project than it looks. Skipping them until you need them was a choice, not an oversight.

  • Payments. Taking money is its own world — tax, refunds, receipts, disputes. Ship something people use for free first. Add payments only when the same person has asked twice.
  • Complex permissions. Teams, roles beyond "owner and admin", per-project sharing. Each new role doubles the number of rules to keep straight. Start with the two roles you already know; add more only when a real person can't do their job without it.
  • A mobile app. Every project in this course opens on a phone browser and works fine. A "real" mobile app (App Store, Play Store) is a separate discipline. Web-on-phone is usually enough for years.
  • Notifications and email at scale. One "the admin gets an email when someone submits" is easy. A full notification system (in-app bells, digests, unsubscribe pages) is not. Delay it.
  • Advanced AI features. Streaming, tool-use, agents, long documents. Lesson 3 used AI in the simplest safe way — as a first-pass helper the human overrides. That pattern carries you a long way.

Part 8 · Glossary refresher

One page. Every term the course used, in the plain-language version you saw the first time. Bookmark this section.

  • Table — a named list of rows inside the database, like a labelled spreadsheet the app can read and write to.
  • Row — one entry inside a table. One volunteer submission, one slide, one tap, one study session.
  • Column — one named piece of info inside a row (the email column, the title column, the date column).
  • Database — the safe place where all the rows live. Lives on an always-on computer, not on any one person's phone.
  • Storage — a separate file locker for big files like images and PDFs. Each file gets a public URL; the database just remembers the URL.
  • Rule (RLS in plain words) — a small instruction that lives inside the database saying "this person is allowed to see/write this row; anyone else is not". The rule can't be bypassed by clever apps or browser tricks — it's checked every single time.
  • Allowlist — a pre-uploaded list of people (usually by email) who are permitted to do something. Anyone not on the list is politely turned away.
  • Auth (sign-in) — the app knows who's using it. Email/password or a Google button. Without auth, the app only knows "someone".
  • Owner-only — a rule that says "only rows stamped with your account id are visible to you". The most common private-app rule.
  • Real time — when one phone changes something, other screens update on their own within a second, without refreshing.
  • CSV / spreadsheet — a plain text file where each line is a row and commas separate the columns. Every spreadsheet app can open one.
  • Slug — the friendly, human-readable tail of a URL, like mei-startup. Chosen once, safe to say out loud.
  • Draft vs published — two states on the same row. Drafts are only visible to the author; published rows are visible to whoever the rule allows.
  • Share preview card — the little tile that appears when a link is pasted into WhatsApp, X, LinkedIn, iMessage, email. Built from two tags on the page (og:title and og:image).
  • Session — a short anonymous id that helps you count "one person hitting refresh 20 times" as one visit. Contains nothing personal.
  • Cloud (the always-on computer) — the machine that hosts your app so it works even when your laptop is closed. Lovable Cloud is one.

Part 9 · Closing note

The point of this course wasn't the five apps. It was the five patterns. Every tool you'll ever build is a remix of these — different words, different faces, same shape underneath.

You now have a mental library: public write / private read (Project A), one power user's private workspace (Project B), many writers sharing one live number (Project C), anonymous group before-vs-after (Project D), and strictly one owner (Project E). When a real problem lands on your desk, name the shape first. The prompt writes itself after that.

Now go find one specific person who's annoyed about one specific thing, and ship them the boring, small, useful tool that removes it. That's the whole game.