Monday kickoff, mid-week behind the scenes, Friday launch — one app, three posts, every week.
German B2B e-invoicing mandate is around the corner. I built a simple ZUGFeRD PDF generator allowing AR clerks to generate a compliant ZUGFeRD PDF as well as AP clerk to validate a ZUGFeRD PDF and get a validation report.
German B2B e-invoicing mandate is around the corner. I built a simple ZUGFeRD PDF generator allowing AR clerks to generate a compliant ZUGFeRD PDF as well as AP clerk to validate a ZUGFeRD PDF and get a validation report.
Providing all the specs for latest ZUGFeRD 2.5 to Claude Code and let it just build this tool was really simple. Especially when specs are cristal clear, AI vibe coding tools do a fantastic job.
On 1st of January 2027, it will become mandatory for suppliers above 800.000€ annual turnover to create EN16931 compliant e-invoices and send them to their customers. ZUGFeRD PDF is one of the recommeded formats here. With this tool AR clerks can easily generate a compliant ZUGFeRD PDF incl. a validation report and even send it via email to their customers. I also build a valdiation tool where AP clerks can upload a ZUGFeRD PDF and proof if it is legally compliant with the ZUGFeRD specs. They can download a validation report as well. Disclaimer: This is not a Basware product. I built this as a private side project.
I build a web app to show the forecast for allergies. Plan is to also launch this as my first mobile app.
I build a web app to show the forecast for allergies. Plan is to also launch this as my first mobile app.
This was the first time I built an app with the mobile phone version of Claude Code. It works well as well, I just don't like that this isn't synched with my Claude Desktop app.
There is a terminal access for the mobile phone, but this requires to keep my Laptop on, which is not really feasible for me.
I have a strong allergy against grass / hay fewer and wanted to build my own small app showing the forecast for allergies.
Invoicia turns dense tax rules into bite-sized lessons, quick quizzes and role-play scenarios. Built for AR clerks, AP clerks and Tax Managers.
Invoicia turns dense tax rules into bite-sized lessons, quick quizzes and role-play scenarios. Built for AR clerks, AP clerks and Tax Managers.
Giving the right context is very crucial to build a correct learning app on a mandate. So I used only trustworthy articles to feed the tools. While cross checking, I didn't find any wrong information which is quite impressive.
Country mandates can be complex to understand, so I built an app to have some fun learning them. Big disclaimer again: This is not a Basware product. I built this on my own as a side project.
This is the first app I built 100% with Claude Code (and actually as well with Lovable).
This is the first app I built 100% with Claude Code (and actually as well with Lovable).
I asked Claude to help me build this.
Prompt:
How can I built a web app with Claude and publish it on a website?
Questions:
Q: What kind of web app are you trying to build?
A: A full app with users, login, databaseQ: How much coding do you want to do yourself?
A: None — Claude does everythingQ: Do you already have somewhere to host it?
A: No, need a free option
And the answer from Claude:
Got it — full app with auth and a database, you don't want to write code, and you need a free host. Here's an honest path that actually works for that combo.
The realistic stack for "Claude does everything"
For a no-code-yourself approach with users, login, and a database, the sweet spot is:
Frontend + backend: Next.js (React framework that handles both UI and server logic in one project)
Auth + database: Supabase or Clerk + Supabase — these give you signup, login, password reset, and a Postgres database without you writing auth code from scratch
Hosting: Vercel (free tier, made by the Next.js team, deploys straight from GitHub)
Code generation: Claude Code (in your terminal) or this chat
All of these have free tiers that get you to a real, live app at a real URL — no credit card required to start.
The honest caveat
"Claude does everything" works well for writing the code. But you'll still need to:
Install a couple of tools on your computer (Node.js, Git, optionally Claude Code)
Create free accounts on GitHub, Vercel, and Supabase
Copy API keys from one dashboard to another
Run a few commands Claude gives you
If that sounds like too much, there are true no-code builders (Bubble, Softr, Lovable, v0) where you describe the app in plain English and they handle hosting too. They're faster to start but less flexible and most have paywalls once you have real users.
Recommended path — step by step
Set up accounts (15 min): GitHub, Vercel, Supabase — all free, sign up with the same email.
Install tools (10 min): Node.js from nodejs.org, Git, and Claude Code (npm install -g @anthropic-ai/claude-code).
Start a Claude Code session in an empty folder and tell it exactly what to build. Example prompt:
"Create a Next.js 14 app with TypeScript and Tailwind. Use Supabase for auth (email + password) and database. The app is [describe what it does]. Set up the database schema, login/signup pages, a protected dashboard, and prepare it for deployment to Vercel."
Hook up Supabase: Claude will tell you which keys to grab from your Supabase dashboard and where to paste them (into a .env.local file).
Push to GitHub: Claude can do this for you with a few commands.
Deploy to Vercel: Import the GitHub repo on Vercel, paste the same env variables, click deploy. You get a live URL in ~2 minutes.
Iterate: Go back to Claude Code and say "add a page that does X" — it edits the code, you push, Vercel redeploys automatically.
Free-tier reality check
Vercel free: plenty for hobby/small apps, ~100GB bandwidth/month
Supabase free: 500MB database, 50K monthly active users, paused after 1 week of inactivity (just hit "unpause")
You can run a real app on this stack with zero monthly cost until you have meaningful traffic
I actually built this app twice: First with Lovable and then with Claude. You can find both via these links: - Lovable: https://family-council.lovable.app/ - Claude: https://family-council-gamma.vercel.app/ I created a poll, so please let me know which one you like more.
They were carved a thousand years ago to protect a castle on a cliff. Tonight they wake on a glass tower above Manhattan — and the city already has enemies.
They were carved a thousand years ago to protect a castle on a cliff. Tonight they wake on a glass tower above Manhattan — and the city already has enemies.
If you don't know what to build next, just ask Lovable in plan mode. I did this after first version was built. This was the reply:
A review of what's built and a menu of 6 improvement directions the user can pick from.
Questions answered
Which direction should I build next?
Plan
You've built a surprisingly complete tactical clan-management game. Here's what's already solid, followed by six directions you could go next.
Title screen, hub navigation, clan roster with day assignments
Full hex-grid tactical combat with 6 story missions, abilities, cooldowns, guard stance, dawn/shelter mechanic
Cutscene system with branching choices that affect loyalty and unlock codex entries
10-gargoyle roster with egg brooding, hatching, and naming
4 stronghold buildings with gameplay benefits
9 random night events between days
Codex/lore gallery with locked undiscovered gargoyles
Cloud saves, auth (email + Google), profile settings with avatar upload
Donation page
1. Sound & atmosphere
The plan mentions ambient loops and combat SFX but nothing is wired yet.
Ambient wind/city drone on hub and title screen
Combat attack sounds, hit reactions, ability stingers
UI click/hover sounds
Optional: a short musical theme for the title
2. Side missions & replayability
Right now the only missions are the 6 story chapters. You could add:
Patrol missions: shorter 3–5 turn skirmishes against random enemy mixes, available from the mission list between chapters
Escort or "survive N turns" win conditions (mentioned in the plan but all current missions are "defeat all")
New Game+ after chapter 6: harder enemies, retained roster, a score tally
3. Combat depth
The combat works but the AI is basic and abilities are under-utilised:
Smarter enemy AI: flanking, retreating to cover, prioritising low-HP targets
Terrain effects: elevated hexes giving +ATK, rubble slowing movement
Combo abilities: two gargoyles adjacent trigger bonus damage
Stone-form timer UI: a clearer countdown to dawn with tension audio
Damage numbers floating on hexes when units are struck
4. Loyalty & character systems
Loyalty is tracked but has no gameplay effect yet:
Low-loyalty gargoyles might refuse assignments or flee combat at low HP
High-loyalty gargoyles gain bonus XP or a "inspired" combat buff
Personal side-quests triggered by loyalty thresholds (e.g. Brooklyn's restless arc, Hudson's honour)
Injury history page per gargoyle showing scars and recoveries
5. More content variety
10–15 more night events (currently 9; repeats get noticeable quickly)
3–4 more enemy types: Quarrymen (from the plan), a boss-tier Steel Clan unit, a fast scout
Cutscene art stills: one per chapter, generated or sourced, shown behind the dialogue
Character detail page at /character/$id with full stats, injury log, and loyalty graph
6. UX polish & onboarding
First-time tutorial overlay on the first mission explaining move, attack, and abilities
Combat tooltips: hover a hex to see if it's in enemy threat range
Better mobile combat controls (the plan says desktop-first, but basic touch friendliness helps)
Mission replay summary: stats screen after victory showing damage dealt, turns taken, gargoyles injured
Settings page game options: volume sliders, toggle combat animations speed
If you want the biggest player-experience lift for the least scope, combine 1 (sound) with 6 (UX polish). If you want more game to play, start with 3 (combat depth) or 2 (side missions).
Pick a direction and I'll break it into buildable steps.
Built a small game with Lovable as a homage to one of my favorite TV series as a kid: Gargoyles
The Arena Where Products Compete to Be #1
The Arena Where Products Compete to Be #1
Idea of the pricing here is to have free of charge viewer users who can vote comment and provide feedback to products and paying users who can promote their products and send their products to battle. The paying users also get AI recommendations how to improve their products, can create product surveys to ask questions and AI summaries all the provided user feedback.
I have chosen a 3 months discounted plan for premium users to make the barrier lower.
I also created a promo code to try it free of charge for the 1st month. Try it out! Promo code is 52apps
You need to enter the promo code on Stripe before actual paying.
Now let's have some fun. Idea of this app is to provide a platform to founders or product owners to send their products into battles. There is 1 champion each week. AI and user feedback helps you improving your products.
Map your invoice format to EN16931: Upload your invoice file and let AI automatically detect the format and map fields to the European e-invoicing norm.
Map your invoice format to EN16931: Upload your invoice file and let AI automatically detect the format and map fields to the European e-invoicing norm.
As AI and LLMs are obviously good in understanding languages, my idea was to let AI figure out the invoice format (just another language?) and map it to the European Norm 16931 to give invoices semantic meaning.
I would say overall the tool does it quite well, but requires more fine-tuning with more complex data structures.
I used the attached SAP IDOC XML as an example which is very simple. Give it a try!
As a user you can still fine-tune the mapping then, save it and export the results in different formats (PDF or JSON).
Built a prototype to map any invoice format with AI to EN16931 (the European invoice norm). You can use this to document how your e-invoice format complies with the European norm (needed e.g. in Germany if you use another format than the 2 recommended ones: XRechnung or ZUGFeRD). Big disclaimer: This is not a Basware product neither a Basware initiative. I implemented and launched this completely as a personal side project.
Small games with AI to learn and have fun
Small games with AI to learn and have fun
Well I used Lovable to generate mini games. Some of them are Ok, but to be honest quite many feel very artificial. So this would require more detailed fine-tuning to improve their quality. At the current stage, I wouldn't let my daughter play these. :D
Built this for my daughter who loves witches and magic and gets to school this year. Idea is to help her prepare for school with games for her age and skills.
Launching this website
Launching this website
Let's start our journey with this website. I want to launch 52 apps in 52 weeks. This is mainly about the journey itself. I want to learn a lot about launching apps and getting your feedback. Maybe there is 1 or 2 ideas worth exploring further. Let's see!