When I was applying for a job earlier this year, part of the challenge was to design a one-page website from scratch. Instead of relying on conventional builders like Wix, it was required to use Lovable. A relatively new AI-powered platform that turns natural language prompts into functional websites (Lovable, 2025). The platform allows users to describe their ideas in plain English, and the system automatically generates a complete layout, style, and structure based on that input (Codecademy, 2025). I had never used an AI website generator before, so I was curious to see whether I could turn a simple concept into a clean, professional-looking site.
Lovable’s free version includes five credits, and every prompt consumes one credit, on average. However, the credit usage depends on how complex the instruction is – some prompts that require more extensive coding or design logic can use multiple credits at once (Lovable, 2025). This constraint made me think carefully about prompt precision, since a vague or overly broad description could quickly exhaust all available credits. I learned that being structured, deliberate, and specific was crucial not only for saving credits but also for producing higher-quality results.
Despite the limitations, I found the creative process surprisingly engaging. It felt like sculpting a digital product step by step – starting with a rough structure and refining it into a polished interface. If I were to do it again, I would approach it with a clearer workflow: first define the basic functionalities of the intended website, then decide on content placement, design and layout, and finally run a short polishing prompt at the end. This method reflects what prompt-engineering research consistently shows: structured, contextual instructions lead to more precise and coherent outputs (Blijleven, 2025; Fagadau et al., 2024).
Another key takeaway was how important human intentionality remains. Before generating anything in Lovable, I used AI chatbots (such as ChatGPT and Claude) to brainstorm the intended website’s purpose, information, and tone (ANTHROPIC PBC, 2025; OpenAI, 2025). This preparation showed me that while AI can accelerate execution, the overall coherence and value still depend on human judgment. I needed to decide on what to include, what to simplify, and how to communicate effectively with the target audience of the intended website.
For our Generative AI assignment for IS, I used Lovable again to prototype our idea. The experience convinced me that tools like Lovable could change web design by lowering technical barriers and increasing creativity. Yet it also raises a broader question: as AI builders continue to improve, will the essence of creativity shift away from coding skills toward prompt design, conceptual thinking, and critical decision-making?
References:
ANTHROPIC PBC. (2025). Overview | Claude. https://claude.com/product/overview
Blijleven, T. (2025, March 6). The importance of a good AI prompt | Spotler. Data Driven Marketing Software | Spotler. https://spotler.com/blog/the-importance-of-a-good-ai-prompt
Codeacademy. (2025). How to create a portfolio website using Lovable AI. Codecademy. https://www.codecademy.com/article/lovable-ai-website-builder
Fagadau, I. D., Mariani, L., Micucci, D., & Riganelli, O. (2024). Analyzing Prompt Influence on Automated Method Generation: An Empirical Study with Copilot. Association for Computing Machinery, 24–34. https://doi.org/10.1145/3643916.3644409
Lovable. (2025). Welcome – lovable documentation. https://docs.lovable.dev/introduction/welcome
OpenAI. (2025). ChatGPT. https://chatgpt.com
I think the question you posed at the end of your post is quite intriguing. The introduction of AI in almost all industries creates the fear of replacement, as AI seems quite disruptive. Especially within the creative fields, like art or website design. However, I believe that AI is still not capable or fleshed out enough to replace creative fields fully. I think AI simply allows people who lack experience and knowledge in a field to catch up more quickly.
Let’s take the example of lovable. The code created by lovable is far from perfect, only creating a prototype that works to some extent. Publishing this prototype, however, is unrealistic, especially if the code is not combed over by someone who understands it fully. One is able to create a layout or prototype of a website/application, but one still requires a systems engineer to create the actual product. There is a creative shift in the sense that prototyping is becoming easier, with users being able to show a systems engineer what they expect from their application more easily.
In summary, AI is still not reliable enough with its outcomes, and therefore, the essence of creativity remains with coding skills. AI can currently only be used as a tool to aid in conceptual designs and prototypes, but it isn’t able to provide a clear, bug-free result that is publishable.
I found your blog very interesting, especially your point about how the limit on prompts changes the way you write them. I’m also using Lovable for the first time, and I find your experiences quite relatable. The credit limit really makes you think about how precise and structured your instructions need to be, which feels quite different from using more traditional website builders. I also agree with your observation that human decision-making still plays an essential role, even with AI tools. It’s a good reminder that effective use of AI still depends on thoughtful human input and a clear sense of direction.