Software that suits you better: this is how it can be done

A tailored workspace 

Erwin Blom, a freelancer in the media and creative industry and writer of the blog FMT Handpicked (opens in new tab), simply had enough: software that is complicated, doesn’t quite do what you want, and suppliers who withdraw a subscription fee every month. So, he decided to do things differently with AI and build something himself. 

Erwin’s ‘Workspace’ (opens in new tab) is a collection of tools that do what he needs in his daily work. From saving sources and notes to tracking projects, tasks, contacts, and invoices. Everything works seamlessly together: a source is linked to a note, which you process in a document, and that document is part of a project with an invoice. 

He stores all his data locally in his own files. The initial idea for the workspace was ready within an hour, while the perfectly refined version took him a month to fine-tune. Because that’s the reality of building software with AI: one prompt doesn’t get you there; the real work lies in refining the details. 

What makes this example particularly interesting is that Erwin made the code public (opens in new tab) (open source). You can use his Workspace for free and adapt it to your own workflow. Missing something? Add it. Too much included? Remove it. 

Erwin's story is a textbook example of vibe coding. In vibe coding, you explain in plain language to an AI what you want to build. The AI then writes the code—the instructions behind the program—and you review, test, and adjust. 

It can go even bigger 

Recently, I also came across another—larger and somewhat wilder—example that shows how quickly this technology is developing and what its possibilities are. The American developer Brandon Thomas had AI build lookalikes of major Adobe and Microsoft programs (opens in new tab). His tool WordCraft closely resembles Word, runs in the browser, and can open and save regular Word files. 

When he shared this with other developers, they responded with great enthusiasm. There’s clearly a hunger for alternatives to expensive software packages. WordCraft currently achieves about 62 percent of what Word can do, so it’s not yet ready for daily use. But the underlying signal is very interesting: software that entire teams worked on for years can now be rebuilt in a short time by one person using AI. 

An important legal caveat: it remains to be seen whether rebuilding existing, copyright-protected software is permissible. No judge has yet ruled on systems like WordCraft. But it does sharply illustrate where the future of software development is heading.

What does this mean for the cultural sector? 

In the cultural sector, the costs of all separate software packages and licenses quickly add up. And many systems don’t work well together. Since custom software is simply unaffordable for many cultural organizations, we often resort to creative (and mostly manual) intermediary solutions. 

Vibe coding offers an interesting alternative: software that does better fit the practices of your cultural organization. 

Rebuilding an entire word processing program might be a step too far, but small, smart custom tools for the cultural sector are already achievable. Want to know what’s already possible? Read our article on vibe coding applications (opens in new tab). 

Prefer to get started right away? In our article on taking the first steps (opens in new tab), we’ll help you on your way. The good news is: if you already know how to give a good prompt to an AI, you’ve got the most important tool at hand. And if you’d rather work on your own tool(s) under guidance, join our workshop ‘AI for the cultural sector—the next step’ (opens in new tab). 

Stay critical: keep your head in the game 

As an AI expert, I see the enormous opportunities of vibe coding, but software doesn’t automatically become free, secure, or flawless. So stay realistic: 

• An initial setup is quick, but refining the details requires real attention. 

• Who manages and fixes the tool if the creator leaves the organization or if an error occurs? 

• AI sometimes writes code with unnoticed security vulnerabilities. Therefore, never process personal data in it without caution.  

• Think carefully in advance about where you host your self-made tool and where the data is stored. 

• For large, critical systems (like your ticketing or cash register), this is not yet an option. 

Nevertheless, there’s a fantastic opportunity here: cultural organizations can once again decide how their digital tools look. Start small, experiment, and stay curious.