The Digital Innovation Industry Course (LTAT.05.019) is a project-based program where four student teams tackle real-world challenges from industry partners each semester. This term, 19 students from diverse academic backgrounds enrolled—some bringing deep technical expertise, while others came from non-technical disciplines. Crucially, this was the first semester where Lovable, an AI-powered application builder, was used for delivering the final prototypes instead of Figma, the more traditional prototyping tool.

For many students, experimenting with Lovable felt like an entirely new frontier in web design and rapid prototyping, offering high-level UX/UI capabilities. The initial experience was described as intuitive and near-magical, particularly for those without extensive coding backgrounds. Students noted that the platform easily surpassed traditional UI tools because it interprets core intent without requiring a rigorous vocabulary of technical jargon. For instance, instead of typing manual commands, students could simply drop in screenshots or mock-ups to generate instant results. One student shared their amazement with this seamless visual process, recalling, “I just copy the UX/UI of any of the websites. Put a screenshot and exactly copy and make it.” For team-based tasks, the speed was unprecedented; another student remarked on how quickly they could develop project ideas, stating, “I was not expecting anything, and three hours later he [the teammate] messaged me and he said that the whole thing is built out.”

The Friction of Real-Time Collaboration

Despite the initial magic, the collaborative workflow highlighted distinct logistical limitations when multiple team members tried to contribute simultaneously. Synchronous development proved difficult because the tool lacks standard architectural checkpoints and a precise version control system. Instead of allowing real-time parallel editing, the software forces prompts into a single processing line.

A student described how they navigated this constraint, explaining, “we were in a team call and then, like, one person made a quick prompt in one thing, another person made a quick prompt in another thing, and during the call, we got our page fixed…” noting that for the bigger changes they had to wait or make different versions of the project. Tracking past edits similarly proved frustrating, with students noting that they had to manually scroll up through a single timeline page as if reading a messenger chat history.

The Challenge of Recovering from Errors

The most prominent hurdle emerged when things went wrong, exposing the difficulties of troubleshooting and recovering from errors within a prompt-based workflow. For those unable to dig into the raw code, breaking the design felt catastrophic because the generated codebase remained completely opaque. As one student expressed, “As soon as I break something, that’s it. Then I lose all my confidence because I can’t go back.” When errors or AI hallucinations cascaded, students had to resort to editing screenshots in Microsoft Paint to physically point out where the layout had broken: “And it just didn’t do what I wanted, so I took a screenshot, edited in paint, gave it the screenshot, and said, this is what I want, and then it worked”.

Even for students with tech-savvy backgrounds—such as software engineers and data scientists—delving under the hood brought its own frustrations. They found that trying to manually edit code directly within Lovable was an incredibly slow, re-rendering process. Upon downloading the source files to diagnose issues locally, they found the architecture to be cluttered and unnatural, noting that it created “so many files, redundant files, and a lot larger than it should have been.”

Why Industry Standards Still Matter

Ultimately, this lack of control convinced many students that it remains absolutely critical to learn classic platforms like Figma rather than relying exclusively on purely prompt-based tools. While Lovable acts like an autonomous agent, students argued that industry design tools teach the necessary fundamental and creative skills required in professional environments. One student pointed out the reality of the job market, emphasizing, “I think that from real content designers, they’re still going to use Figma for years and years to come…” Another student echoed: “Actually, in the real world, Figma is used.”

They noted that a software engineer who cannot navigate design software will be at a major disadvantage when entering a startup or consulting with clients. Furthermore, Figma was praised for offering a hybrid flexibility that AI-only tools cannot provide. As another student concluded, “I think Figma has the sweet spot there that you can do both. You can use AI and you can do it with your own hands. Lovable is only whiteboarding.”

Looking Ahead

From a course deliverables perspective, Lovable allowed students to deliver more detailed and functional solutions for industry partners in a much faster manner. While the students’ overall experience was highly positive, the semester revealed a critical takeaway: jumping straight into AI prototyping can be a double-edged sword, occasionally risking the neglect of essential technical skills.