I wanted to take this AI vibe coding adventure a step further.

Remember my Setlistplanner.com idea? So far - after 2 years - it has only one paid subscriber. Long story short: it's a failure. By the way, I'm 99% sure the same is going to happen with FlowDent.be. I'm not even going to make enough money with it to pay for the 2 domains I bought for it.

So I asked AI to rebuild SetlistPlanner. But this time I wanted a ZERO-cost hosted solution. No monthly payments for hosting, storage, database, or backups. And no advertising in my app or on the landing page.

This came out of it: https://ai.setlistplanner.com/ (I can even add my own domain, it's free!)

Opus 4.6 built this in minutes. Cost: $17.24.

I did not specify any technical details and let Opus decide on the stacks. Opus chose the following and explained why each time:

Just to make things clear: I don't know, nor have I used, any of the above technologies. I'm a good old Classic ASP guy, remember... I do know one thing though. Using frameworks like React brings my "development skills" to 2026. I can now deploy applications that do not need "refreshes" to get the latest updates. This is very interesting in the case of a setlist planner. When I play with the band, I often decide at the last minute to skip a song, add one on the fly, or change the sort order. That makes my band members crazy. They use tablets during a live performance. Now I can be sure that when I do so, their setlists get changed without any hard refresh (F5). 

Another thing. The time between me prompting for a change or a bug fix - or why not, a new module - and the deployment on the live site is only minutes, sometimes less. I prompt for a code change in OpenCode, Opus 4.6 takes care of it, and it automatically pushes the changes to the GitHub repository (through Shell). Seconds later, Vercel automatically deploys the changes from the GitHub repo. Just like that. No manual uploads. No human interventions. I can literally speak to OpenCode and be sure that minutes later a new module or a bug fix is deployed. 

So far so good. However...

I need the following accounts to get this done:

1. I'm using OpenCode.ai. I need an account there, as I only use paid AI, mainly Opus 4.6.
2. I need an account at Claude to create the necessary API keys for OpenCode to use.
3. I need a GitHub account where the code repo is stored.
4. I need a Supabase account to deploy databases.
5. I need a Vercel account to deploy the codebase (so Vercel needs to be linked up to GitHub).
6. Actually, I've used my Gmail account just about everywhere I could. Makes life easier.

So this little app requires 6 accounts for 6 independent cloud-based services (OpenCode (Zen), Claude (Anthropic), Google, GitHub, Supabase, Vercel). That's a lot of dependencies. How sure can I be that in 2 years from now, this little web application will still be available? And for free? It's a long shot.

On the other hand, once you get used to this complex architecture... you can deploy dozens of such SaaS-based apps a day. There are no limits. And the only company that makes money out of this, is Anthropic (Claude). Maybe that's why Anthropic has crashed the software stock market.

To be continued.

I was pretty quiet these past few days, wasn't I? And if I know me, something nice was going to come along: https://flowdent.be/ - built on ASPPY. It uses PDF generation, json, zip, bcrypt, jpg-resizing... in other words, everything I needed ASPPY for in the first place.

FlowDent is a complete, free tool for dental practice management: patients, appointments, stock, digital signatures and more. No hidden fees, no limits.

Spread the news!

Another morning, another day-after feeling that already makes my day.

Yesterday, Opus 4.6 created yet another killer web app (it's login based, so not much to see) for me. My wife needed a digital sign-off application for her dental practice. From one day to the next - in Belgium - dentists now have to let patients explicitly agree to the cost of each and every dental treatment. Patients must sign a document clearly explaining the pricing and planning of any treatment.

So all of a sudden, tens of thousands of dentists now need a solution to facilitate that. To avoid a lot of paper printing (and scanning), a digital sign-off solution has become extremely useful. Not just for signatures, but also to provide templates and enable near-100% automation of this process.

There are already some existing SaaS-based solutions. Still, Opus 4.6 built me a wonderful little web app that takes care of everything. And it looks pretty good too. 80 USD in AI credits. Done.

I probably shouldn’t share my ideas here. But what if I ask Opus to turn this into a multi-user, multi-language, multi-everything web app that dentists can start using within minutes - not even hours? Add a monthly or yearly fee. Beat the competition with ease. And start selling tomorrow? Well… I can.

SaaS providers are going to face a very difficult time in the coming years. New SaaS models will pop up - and disappear - just as quickly. With every new major AI release, new opportunities will emerge. Software will be easily rewritten and upgraded using the latest AI models time and again. Large software companies will NEVER be able to keep up with this. Solo developers will, with ease. 

From basic idea to producion-ready to killer web app in half a Sunday: LifeAdmin.be

Too many features and use cases to even imagine. The autogenerated website and blogpost tries anyway.

80 USD on AI credits using Claude Opus 4.6. Multiuser, multilanguage, multi-everything. I haven't written a single line of code myself. Runs on ASPPY. MVC. Load times below 100ms. This is something I wanted to do for a while. And I did it, in half a day. Would have taken me 6 months to develop this me handcoding.

It's a reference kind of app. You can use it or give it a try, but chances are big that I need to wipe all data at some point. I'm gonna use it myself in the first place. And filter out the bugs in the coming days.

Claude Opus 4.6 created a basic CRM in under 10 minutes. Cost: $ 5.31.

My prompt? One line: Create a basic CRM with companies, contacts and leads. I pasted that prompt in the ASPPY promptbuilder and fired up Opencode. Next I pasted the JSON-prompt from the prompt builder into a fresh (no-context!) Opencode session. Opus inspected the prompt, read the docs, studied my ASPPY framework, and started off. 10 minutes later, a complete MVC-driven, secure, completely tested and ready to deploy CRM was born.

So wait...

• This runs on MY OWN AI-generated runtime (ASPPY)
• No IIS, asp.dll or VBScript dependencies. No license fees. Runs on Linux, Windows and macOS.
• It's MY codebase, I OWN the code
• It's developed on MY PC, not in the cloud.
• And MY runtime ASPPY is available - for free - on Github and can be read/understood/improved by ANY AI model in minutes. I AM NOT NEEDED ANYMORE! I'll die in peace now.
• And it's MY 5.31 USD. Or no wait, that's the bad news.

The era of the solo developer has arrived. Actually, it's back. Between 1995 and 2005, solo developers made the internet great. Thanks to AI, we're back! But this time, we're there to stay. We will develop our own operating systems, our own browsers, our own programming languages, our own application runtimes and our own frameworks. No dependencies. No one to rely on. No borders. No limits. Freedom.

Now I'm able to even improve ASPPY based on some mistakes I saw Opus make: Request.Method? That did not exist in ASP 3.0 either! Classic ASP had Request.ServerVariables("REQUEST_METHOD").

ASP.NET: uses Request.HttpMethod. Python (Flask/Django): Uses request.method.

That's where this error came from. So guess what? Both methods are now supported in the Request-class in ASPPY. No more guessing. No more waste of time and money. It's in there and from now on request.method just works in ASPPY. 

I repeat. Sell. Your. Software. Shares. They will go nothing but down from now on.