From tools to workflows

13

October

2025

No ratings yet.

As a strategic designer working at the intersection of technology and human-centered design, I experienced first-hand how generative AI is reshaping the traditional design process. During the Tech-Enabled Innovation Studio course at TU Delft, my team and I developed an AI-driven triage assistant for Dutch primary care. This course was a testing ground for exploring how tools such like Lovable, NotebookLM and Cursor could influence creativity and bridge technical and strategic thinking. By combining these tools, it became possible to create feedback loops across design stages, for instance, using ChatGPT to refine prompts for Lovable or iteratively improving Lovable’s chatbot functionality through Cursor’s code environment. Eventually, combining them as interconnected components within a data-driven workflow.

Hatchuel and Weil (2003) introduced C–K Theory as a new approach to innovative design. In this framework, the Concept Space (C-space) contains undeveloped ideas and hypotheses, while the Knowledge Space (K-space) holds established information and validated knowledge. I believe AI’s greatest strength lays in its ability to move quickly between Concept and Knowledge space. Within minutes, abstract prompts turned into tangible prototypes that could be tested and refined live with stakeholders like General Practitioners. Especially Loveable was super powerful in translating our ideas into artefacts. From a BIM perspective, this mirrors the dual process of data exploration and knowledge forming, central to information management. This allowed me to focus more on strategic decision-making and less on technical constraints, ultimately improving the collaborative and reflective nature of the design process.

Yet, the experimentation also revealed limitations. AI outputs often lacked depth or consistency after initial iterations, making it very difficult to maintain quality. Synthetic users, AI-generated representations of interview participants, were helpful for quick feedback but could never replace real human nuance. Moreover, the environmental footprint of continuous AI use added an ethical dimension to the conversation about responsible design and development.

This experience underscored that the future of digital innovation depends less on individual AI tools and more on the integration of intelligent systems within coherent information (work)flows. Generative AI can strengthen decision-making when combined with strong information architectures that ensure traceability, context, and human oversight. The role of the strategic designer or BIM professional, therefore, is to act as the leader and guider of these hybrid ecosystems, defining what AI should contribute to, aligning it with organisational goals and maintaining the balance between automation and human judgment.

Sources

Hatchuel, A., & Weil, B. (2003). A new approach of innovative design: An introduction to C-K theory. In Proceedings of the International Conference on Engineering Design (ICED 03), Stockholm, 19-21 August 2003. 

Please rate this

Using GenAI the ENCOM Way

13

October

2025

No ratings yet.

As a die-hard Tron fan, I recently went to watch Tron Ares in the cinema, and it stuck with me.  In the Tron world, ENCOM and Dillinger Systems are rival companies competing for control of the digital future. Without spoiling the whole story, I can say that the movie shows two paths for AI: the ENCOM path, where tech helps people and opens new possibilities to improve our lives, and the Dillinger path, where power and control come first. Seeing that contrast on a giant screen made me think about how I use generative AI in my own life and why my experience has been mostly positive.

Day-to-day, GenAI, mostly large language models (AI chatbots), feels like an ENCOM tool for me. I use it to give me medical advice, clean up emails and messages, translate conversations. It summarizes long articles when I’m short on time, helps me outline essays, drafts slides and talking points. When I’m coding with RStudio, it explains errors in simple terms. And when I’m planning a trip, it helps me think through checklists. It’s like an assistant that only wants the best for you.

Part of why this stays positive is the way these GenAI tools are set up. It tries to be friendly and helpful, and it puts safety first. Harmful or abusive requests get blocked or redirected. It won’t imitate a living person’s exact voice, won’t help with dangerous instructions, and pushes me toward responsible use. That doesn’t make misuse impossible, but it does make it harder.

Of course, the Dillinger path exists in the real world, too. We see AI being built into defense, border security, and large-scale surveillance systems. Companies like Palantir and Anduril are known for powerful analytics and autonomous sensing platforms. Facial recognition firms have scraped massive image datasets. These tools can centralize power in ways that can be worrying. It’s very much like the movie’s warning: when a few actors control the Grid, ordinary Users lose agency. I’m not saying these companies are “villains,” but the direction of travel still matters.

So I set myself a simple goal: keep my use of AI on the ENCOM path. Tron Ares should remind us that AI Programs can turn dangerously powerful in the future. If we give them good goals, they can light up the city. If we don’t, the same power can bite back, with the risk of even turning against us. Kind of like Skynet, but that’s a whole other franchise.

Please rate this

The coding revolution called “Cursor”

12

October

2025

No ratings yet.

During the last two weeks, for one of my courses, I have been experimenting with Cursor.
Cursor is a generative AI powered coding tool. So basically, it allows you to do software
coding without actually having to know anything about coding. Unlike the traditional code
programs like R, Cursor can actually understand natural language and can generate the
code based off of the natural language input it gets from a person. Basically, it feels like
having a senior coding expert writing code from our input real time with a 24/7 availability.
When I first used it, Cursor seemed a bit strange and I was not sure how it worked. However,
it only took me a couple of hours in total to go from a complete beginner to trying to generate
my own app idea. The AI lets a user write their ideas and it generates the complex code for
them in real time. After a few hours, the base of my idea for a virtual fitting room was already
there and it looked amazing (Figure 1).


Figure 1


I was super happy with the result that I got and even though it is of course far from a finished
app I was extremely impressed with Cursor this far. However, my experience with Cursor
also showed me some areas where I think they can improve. For example, while most of
Cursor’s suggestions are good, it sometimes also overlooks some logic steps or some
security nuances. With future updates, I would like for a future version to have a better
project memory. Right now, it can sometimes feel as if the application needs you to provide
the full idea and some of the progress again and again in each prompt.


Overall, Cursor has transformed the world of coding, making it more accessible to
everybody

Please rate this

Ask first or attempt first?

12

October

2025

No ratings yet.
Source: Shahi, S. (2024, May 1). How to balance human Creativity and Generative AI in product Marketing. Innovation & Tech Today. https://innotechtoday.com/how-to-balance-human-creativity-and-generative-ai-in-product-marketing/

Just like you, as a student in 2025, it is probably not surprising that I use genAI tools, such as ChatGPT, almost every day. What has been almost stigmatized in university settings in late 2022, now is an open discussion and a core part of learning and student life. Personally, I use ChatGPT regularly for school work which, I believe, is pretty similar to what my fellow students do. Just to mention a few instances: brainstorming for projects, cleaning up messy notes into structured study summaries, getting quick explanations for terms I don’t fully get, and even asking it to “grade” my drafts against the rubric so I know what to fix. One of my favourite practice, tho, is to feed the lecture notes to ChatGPT and ask for a 10-15 minute podcast of the key takeaways and learning goals. I listen to this during my walk to school and feel like I’ve revised without sitting behind my laptop. It’s fast, helpful, and honestly makes me feel on top of things when I’m at the lecture.

I also use GenAI outside of school. For example, when my parents visited me, I asked it to plan a simple Netherlands tourist route for city trips so they could see the highlights without me stressing over the details. Another instance is that on random weeknights, I’ll throw in three or four ingredients I have at home and get a quick dinner idea that actually works. This saves me a lot of time and money usually, and I also get to try new recipes. However, after our guest lecture on GenAI tools, it really hit me that there are tons of other options out there, but I keep defaulting to ChatGPT. It has really became my comfort zone when it comes to using AI.

But here are some contradictory feelings. I’ve started noticing a reflex: the moment I hit a tough problem, I want to ask the bot before I even try. When I do that too often, my brain feels a bit “sleepy”. Like I’m outsourcing or giving away the hard, creative part. Funnily enough, when I otherwise force myself to work without it, a different feeling kicks in: what if I’m falling behind because everyone else is using AI and shipping better, faster? That AI “FOMO” is real, especially when our standards are also rising with the AI-powered solutions. Anyone else feel that?

As a last thought, sometimes I think about kids, the next generation, who’ll never remember school without GenAI. Lucky them? Maybe. I’m half jealous of the time they’ll save and half protective of the slow, messy struggle that taught me to think. What do we want their first reflex to be: ask or attempt?

Please rate this

Color Analysis, Pictures & GenAI

11

October

2025

No ratings yet.

These years, a beauty test has become popular in South Korea and then in east Asia, which is called color analysis. Color analysis is a method of image management that classifies people into four seasonal types—spring, summer, autumn, and winter. At first, beauty clinic started with physical colorful clothes. They offer clothes of various clothes and make customers wear them. By comparing and analyzing the visual effect that colors combined the skin tone, hair color, and eye color, the customers will get personal style suggestions. For example, individuals with a warm-toned spring type suit bright yellows and oranges, while those with a cool-toned winter type are better suited to highly saturated blues and reds.

Such a color test in offline beauty clinic costs around 100 euro one time. However, with the help of GenAI, we can do it for free. I took a selfie under natural sunlight, without makeup or filter, then used the edit function in my phone to get the RGB value of my hair color, skin color and eye color. After that, I sent those RGB values to ChatGPT and asked it to give me advice following the format of color analysis. Finally I got a detailed report on beauty suggestions like clothes style and makeup style, saying I was the deep autumn type, which turned out to fit me very well.

Another GenAI function I use usually is generating pictures. My part time job is about managing a social media so I need to post many pictures according to the text. I tell ChatGPT to make a poster about a webinar, with captions of its theme, time and location. Unfortunately, in most cases the picture is not what I imagined, even I gave a clear instruction. I usually had to make adjustments for four or five times before getting a just-so-so pic. At first, I guessed the free version was the one to blame and maybe a premium could do better. But after a month of premium version I gave up. The picture generation is just not so satisfying as text analyzing function.

I believe if I give a descriptive instruction following the coding format, the picture it generates could be perfect. But the problem is that I use GenAI tool to save time and energy. If I have to type the programming code for a picture, why don’t I just make a new picture in Photoshop? In my opinion, GenAI tools are already quite smart about analysis and reactions to very clear quantitative orders. More improvements could be done in more abstract and more practical areas, especially under qualitative instructions. For example, when I say “zip up the character’s jacket, keeping the rest of the image unchanged”, don’t generate a picture where the person in it has two right eyes.

Please rate this

How AI Can Turn Ideas Into Organized Maps

10

October

2025

No ratings yet.

We often use mind maps to organize our thoughts, connect ideas, and visualize complex topics in a simple but structured way. They’re especially useful for brainstorming, note-taking or preparing projects. However, it can take time to create a detailed mind map. Especially when we have to summarize long articles, videos or research papers. That’s where AI tools like Mapify.so come in. A tool like Mapify makes it easier and faster to create these clear mind maps automatically, helping us save time.

After signing up for Mapify.so, the platform asks how you plan to use it: for work, education or personal projects. It then gives a short introduction about its features. Some key features include:

  • Free on-page summarization
  • Suppport for 30+ languages
  • Works with Youtube videos, PDFs, documents, research papers, blog posts, social media posts and even emails
  • You can ask questions and the tool will research and brainstorm ideas for you
  • There are two AI models: Instant (faster, credit-efficient) and Powerful (more detailed, higher quality)
  • Option to generate mind maps step by step for more control or all at once for quick results
  • And lastly a wide selection of templates for brainstorming, outlining, project planning, analysis and more

For my test, I uploaded a Youtube video essay called “Drowning in Entertainment: The Age of Distraction.” Within minutes, Mapify analyzed the 33-minute video and turned it into a color-coded mind map that summarized the entire discussion clearly. Each subtopic had its own color, making the structure easy to follow and visually appealing. I could also change the format (logic chart, tree chart, timeline or standard mind map) and present it as slides. One of the most impressive features was being able to view the transcript of the video and even chat with the AI tool about the content. 

Essentially, I was able to understand a 33-minute video in under 5 minutes, without even watching it. When finished, I could export or share the mind map and even tag it for better organization in my account.

Besides the free plan (which gives you 10 credits), Mapify offers 3 paid options: 

  • Basic Plan: €5.99/month
  • Pro Plan: €11.99/month
  • Unlimited: €17.99/month

You can receive a 40% discount if you take a year plan. The basic plan excludes audio summarization and image uploads, while the pro plan adds text-to-image generation and more file types. Personally, I’d choose the pro plan, as it offers the best balance between price and functionality. The unlimited plan mainly includes early access and premium perks that I don’t find essential.

Overall, Mapify.so is a powerful, time-saving AI tool for students, professionals and anyone who wants to learn efficiently. However, I did face some navigation challenges. Improving the user interface with a clearer navigation bar or dashboard could make the experience smoother.
Despite these small issues, Mapify.so stands out as an efficient AI tool for summarizing and visualizing information. It’s an excellent example of how generative AI can make productivity more effective.

Mapify: AI Mind Map Summarizer. (n.d.-b). Mapify. https://mapify.so/

Please rate this

My Personal Experience using GenAI

10

October

2025

No ratings yet.

As a Master’s student who also works part-time as a software engineer, I’ve been relying more and more on generative AI tools in my daily life. At first I treated them as a novelty, but over time they’ve become a multipurpose tool for both my academic and professional life. Still, GenAI is not a magical tool that solves everything, its limitations have become increasingly clear to me over time.

One of the earliest ways I used GenAI was for search. Instead of digging through ten different Google results, I could ask a direct question and get a straight answer. This saved time when I was researching for papers, looking up a software library, or just wanting an answer to random questions I had (as one does). However, I was cautioned against trusting the answers blindly. I quickly experienced first hand that sometimes the AI gives outdated information or confidently states something incorrect. So I do additional research, depending on how high the stakes are.

Another major use case for me is summarizing. During my studies I often have to digest long articles, papers, or lecture notes. Letting the AI condense 20 pages into a quick summary was a game changer. Of course I still have to do more in-depth reading when I need to fully understand an argument, but it gives me a head start and helps me prioritize what to focus on.

GenAI is quite good at brainstorming and drafting too. Whether for a group project in class or when sketching ideas for a feature at work, it provided prompts and perspectives I wouldn’t have come up with myself. The downside is that its creativity can be surface-level, models often just regurgitate variations of ideas that were in their training data. So if I want to come up with something truly novel, I try to think of it myself and then use GenAI for “validation”.

In terms of drafting, I’ve used it to outline essays, emails, and even software documentation for my work. It’s great for overcoming writer’s block and speeding up the initial phase. Still, if I don’t rewrite and refine the draft myself, it’s easy to see that it was generated by AI because it sounds generic.

In my work, I use GenAI mostly for boilerplate code and bug explanations. It saves me time on repetitive tasks. But in complex systems, its contextual capabilities fall short. I’ve had it produce code that looked correct but had subtle flaws, not in syntax but how it used other functions in the codebase. Also, at times it has difficulty adhering to the design philosophy and stylistic choices of larger projects.

Finally, I’ve even used GenAI for language learning (currently Dutch). It’s particularly good for practicing small conversations and checking grammar. That said, I heard from a couple Dutch friends that it sometimes uses phrases that feel unnatural to native speakers. However, for now its Dutch is definitely better than mine, so I will continue using it for learning.

In sum, I already use GenAI for a variety of tasks, and I’m sure I will continue to discover new ways it can be useful. Have you tried any of the use cases I mentioned? What was your experience? I’m curious to hear.

Please rate this

When Football meets NFTs

10

October

2025

No ratings yet.

In recent years, NFTs, also known as non-fungible tokens, have taken the digital world by storm. Simply put, an NFT is a one-of-a-kind digital asset that proves ownership of something unique, like an image, video, or collectible. In football, they’ve been promoted as a new way for fans to connect with players, buy digital player cards, or access special experiences (Sorare, n.d.).

But as exciting as it sounds, not every project is as trustworthy as it appears. A recent case in Spain shows how easily an emerging technology like this can turn into controversy.

According to PANews, authorities in Barcelona are currently investigating a cryptocurrency fraud case involving several well-known football players. The company Shirtum allegedly scammed investors out of about $3.4 million through an NFT project that never delivered what it promised. Some big names were tied to the project: Papu Gómez, Lucas Ocampos, Ivan Rakitic, Javier Saviola, Nico Pareja and Alberto Moreno. Prosecutors claim that Shirtum operated through a complicated business structure to avoid paying taxes. The players reportedly helped promote the NFTs as “founders.” These tokens were meant to represent the players’ image rights, but investigators say they never gained any real trading function. After the promotion, related posts on social media disappeared and the company later claimed it had been hacked without filing a police report. Soon after, Shirtum shut down completely without any explanation (Binance, 2025).

As someone who enjoys watching football, I believe NFTs have the potential to make the sport more interactive if used responsibly. Unfortunately, scandals like Shirtum’s make fans lose trust. For NFTs to succeed in sports, transparency should come first. Football players and clubs need to ensure that what they promote is legitimate and provides genuine benefits. In the end, NFTs should be about strengthening fan communities, not taking advantage of the loyalty and passion of fans.


Sources:

Binance. (2025, June 11). Barcelona court investigates cryptocurrency fraud involving football players. Binance. https://www.binance.com/en/square/post/06-11-2025-barcelona-court-investigates-cryptocurrency-fraud-involving-football-players-25461462914385

Sorare. (n.d.). https://sorare.com/football

Please rate this

When Algorithms Meet Aesthetics

10

October

2025

No ratings yet.

Have you ever wondered how your room or house could look better? What if you rearranged your furniture, changed the floor, or added a new color to the walls?

In my case, I wasn’t sure what to do with a particular part of my room. About a year ago I moved, and my new room includes a staircase leading to an upper area. I wasn’t sure how to use that space, so I sent pictures of my room to ChatGPT and Microsoft Copilot with a short prompt describing the style I had in mind. Then these AI tools gave me several ideas like placing my bed upstairs, creating a cozy lounge with a couch, or turning it into a walk-in closet.

In addition, I asked ChatGPT and Copilot to generate visualizations of that space, so I could decide more easily. It was impressive to see how that space could be used, and how specific styles like Metropolitan Luxury or Japandi could be applied. This makes interior styling much easier because you already have a clear idea of what you want. Instead of making all the changes and realizing afterwards that you don’t like the result, you can visualize it first. It saves time, effort and money.

However, there is still room for improvement for this feature of Copilot and ChatGPT. These tools don’t always get the room layout completely right, so sometimes the window would be on the wrong side or the shape of the ceiling wasn’t accurately represented. The generated images can sometimes also look a bit unrealistic or cartoon-like. Even so, when you upload inspiration photos as well, the results become better and more realistic. Besides, when you ask specific edits it does not necessarily change the result how you would want. Perhaps other generative AI tools that are specifically designed for interior design perform better.

Overall, using AI for interior design is a nice way to get inspiration and explore ideas quicky. Would you use AI to design or redecorate your room?  

Bibliography

Microsoft Copilot. (2025). AI-generated image [Generative AI output]. Microsoft. https://copilot.microsoft.com

Please rate this

AI : A Colourful Time Machine

10

October

2025

No ratings yet.

I believe it was a year ago that I found image restoration models became so easily available. As a kid, I used to watch National Geographic and the Smithsonian Channel where they aired colorised historical footages, which required countless hours of work by teams of colorists, archivers and restoration units to produce. Today, it is just a JPG drop and prompt away.

Considering what these tools are capable of nowadays, I know it was a small experiment, yet I found some interesting insights to share. I wanted to see if I could bring colour and clarity back to old photographs of my grandfather, Catello Spagnuolo Sr., and his brothers. They were all young, smiling, and full of life, but in a greyed out world. I uploaded one of their group photos into a visual restoration tool and wrote a simple prompt: Restore this image, fix tears & dust, enhance contrast, colorize gently.

I utilised two different AI tools: Gemini 1.5 Pro and DALL·E 3. Gemini 1.5 Pro impressed me with how naturally it handled portraits. In close-ups, it maintained the context of light and natural skin tones, revealing subtle details like the warmth in my grandfather’s face, making his eyes shine. But when it came to group photos, it struggled to identify the characters, resulting in a bust with a dash of blue.

DALL·E 3, on the other hand, left me speechless. It lit a thousand colours in the picture and made it feel alive. However, these colours for their suits, shirts, and ties were entirely imagined. Yet the model was able to maintain minute details like the striped fantasy of my grandfather tie, again impressive!

I felt captivated by the hallucinated colors AI decided to use to dress these men.

Seeing my grandfather’s in its youth, its expression in color made me feel closer to that time and the man. It remains a frozen moment, but a bit warmer.

AI didn’t just restore old photographs; it created a connection for me. If you find any pictures around in black and white, I urge you all to try it out.

Please rate this