No ratings yet.

Today’s fashion retail world center is undoubtedly online, and although online shopping is surely convenient, its convenience comes with complexity. It is still significantly difficult for buyers to find products that perfectly match their style and preferences, with discovery fatigue being a frequent feeling that online shoppers face. Since the current search and filtering tools being used rely heavily on keywords, they often find it frustrating not being able to express what one truly wants (Ding et al, 2023). Additionally, even though personalization is undoubtedly a core expectation of buyers (especially the younger generations), e-shops still seem to focus on suggestions, up-selling and cross-selling rather than understanding each client’s intent and aesthetics (Cheng et al., 2021).

This inspired Vestia, a GenAI-based assistant that reinvents the online shopping journey by transforming it into a customized, human-like styling experience. Instead of just typing or choosing keywords, online shoppers describe what they want in natural language (for example “smart-casual outfit for a Friday dinner under €100”) and Vestia instantly generates complete outfits tailored to their style, size, and budget. 

Since the shopping assistant is integrated directly into each retailer’s e-shop via modular APIs, Vestia acts as a seamless module that enhances the “evaluate and choose” step of shopping, rather than replacing the storefront of the website. This way, retailers keep full control over their catalog information and branding aspects, while their potential customers enjoy a seamless yet enhanced experience.

Vestia’s intelligence, relevance and accuracy in understanding styling preferences and generating valid suggestions are powered by continuous learning from multimodal inputs, as well as from user feedback in the form of clicks, skips, and purchases. This dynamic personalization benefits both shoppers and retailers. Shoppers remain less stressed while browsing through endless products and make confident decisions about entire outfits without worrying about having to return their purchases. As for retailers, pointing customers to more relevant complementary and alternative items increases their conversion rates, average order values, and operational efficiency, as well as improves product exposure and overall sales distribution.

Access to Vestia’s shopping assistant is available to retailers in three different packages (basic, pro, and enterprise), depending on the features and support levels they need. By paying, retailers also gain access to service level agreements, as well as to an analytics dashboard that provides them with actionable insights, including outfit views, add-to-cart rates, popular styles, and peak usage times, enabling informed merchandising and marketing decisions. For end-users, Vestia’s shopping experience is for free, through the retailer that has enabled it. With this pricing structure, it is possible to scale across retailers of all sizes and match value to usage and results.

Vestia transforms online fashion shopping by turning it into a more personalized, user-friendly, and engaging experience. To experience Vestia in action, you can visit our prototype here: https://vestia-style-guide.lovable.app 

Team 19

---

References

Ding, Y., Lai, C. S., Mok, P. Y., & Chua, T.-S. (2023a). Computational technologies for fashion recommendation: A survey. ACM Computing Surveys. https://doi.org/10.1145/3627100

Ding, Y., Lai, C. S., Mok, P. Y., & Chua, T.-S. (2023b). Fashion recommender systems: A review and future outlook. arXiv. https://arxiv.org/abs/2306.03395

Cheng, Z., Chang, X., Zhu, L., & Kankanhalli, M. (2021). Fashion recommendation and complementary outfit retrieval: A survey. ACM Computing Surveys, 54(5), 1–38.

No ratings yet.

In general, brainstorming with a machine feels odd, yet oddly familiar. It’s like having a teammate who never stops coming up with new ideas. With any type of prompt, no matter how odd, short, generic or ambiguous, Generative AI will produce something in response, with a reply that can be either brilliant or just a stream of cliché ideas.

Its speed is the first thing to catch your interest, being almost intoxicating. The moment you enter your initial idea, AI already has ten different suggestions ready for you, so you skip the tiresome step of idea formation. But then comes the real work: deciding which ideas have depth and which are just noise. Just because AI can tirelessly generate ideas and concepts, it doesn’t mean that it knows the difference between a meaningful thought and a mix of gibberish, meaning that that responsibility still falls to you.

If you observe AI’s delivery of ideas, the contrast between human judgement and plain generation of content becomes even more noticeable. When working with a team of human members, the “failed” ideas are the ones that keep the conversation running. Even though they can lead to intense discussions, these ideas spark debates and guide the team to better and stronger outcomes, assisting the brainstorming session to achieve the best possible result. AI, by contrast, just keeps producing, indifferent to whether it strikes gold or nonsense. And if you try to correct it over an absurd suggestion, it just accepts your criticism and moves on, without even remotely debating your thought process.

This lack of debate can have both advantages and disadvantages. On one hand, there is a feeling of freedom when you can explore your ideas freely, without getting interrupted or being judged. On the other hand, that makes you the only person responsible for recognizing value, and for every refinement, adjustment and decision needed. AI acts as an idea enhancer, and its concepts don’t amount to a breakthrough unless you take the time to consider them.

What I’ve found through using AI is that handling it as a conversation partner makes sessions more impactful. If I keep asking, refining and challenging its outputs, it ends up expanding my imagination rather than replacing it, as it sometimes reflects possibilities I haven’t yet considered, making me rethink assumptions and spot new directions to follow. Although AI can endlessly generate new ideas, only its human users can turn them into something truly perceptive. However, that leaves me wondering: As machines are being used increasingly more to amplify our thinking, how do we define the line between inspiration and invention?

No ratings yet.

The courtroom of the future is digital, challenging many centuries of legal tradition via automated systems with artificial intelligence (AI). Automated systems, together referred to as LegalTech, now provide legal services beyond the direct involvement of qualified legal teams. Clients now find said services easier and faster, and likely more affordably. This all results from available tools streamlining complicated legal procedures. These tools include document preparation, case outcome prediction, and also large-scale data analysis (Branovitskii, 2025).

Work that previously required hours of human labor by law experts is becoming more efficient due to streamlining by AI tools. Platforms like Kira Systems and Luminance can review and analyze key data from contracts and documents at scale, significantly reducing the time needed for routine legal tasks (Kira Systems, n.d.; Luminance, n.d.). This changes how lawyers operate, allowing them to focus on their more complex tasks at hand rather than spend their time on standardized procedures.

Another area of legal technology involves the use of online dispute resolution (ODR) platforms. Agreements may be enforced automatically by smart contracts that blockchain technology drives, as platforms like Modria ease the digital resolution of disputes. This specific platform manages case progression, suggests next steps, and sometimes proposes settlements through algorithms and rule-based systems (Modria, n.d.).

These innovations make things faster and simpler, however they also raise ethical and legal concerns, particularly related to the algorithmic decision-making and error responsibility. Artificial intelligence, no matter how advanced, could be inadequate for the situations requiring factual conditions that are particular. In complex situations, standardized algorithms may miss important details. In LegalTech’s case, they may overlook facts of legal significance that appear insignificant, and they may even “invent” examples that don’t exist in order to reinforce their claims. Real examples of such cases exist. A New York attorney who used ChatGPT as a means to find precedent cases ended up facing a court hearing of his own, as the AI tool fabricated six non-existent cases (Armstrong, 2023). The tools mentioned above can quickly and reliably handle large amounts of routine work, but they cannot completely replace the judgment and analysis that attorneys provide. Efficiency must never come at the cost of justice, ethics, or professional responsibility.

LegalTech is here to stay. Technological innovation must be balanced with human judgment, so lawyers need to remain central to interpret, guide, and safeguard the law. Yet, as AI and automation take on more routine legal work, the question that remains is: How can society be sure that efficiency in legal services does not sacrifice justice, ethics, and the advice that only humans provide?

References

Armstrong, K. (2023, May 28). ChatGPT: US lawyer admits using AI for case research. BBC. https://www.bbc.com/news/world-us-canada-65735769

Branovitskii, K. L. (2025). Artificial intelligence and LegalTech: Risks of transforming the legal profession. Digital Law Journal.

Kira Systems. (n.d.). AI contract review software. Retrieved from https://www.litera.com/products/kira

Luminance. (n.d.). Legal-Grade™ AI. Retrieved from https://www.luminance.com/

Modria. (n.d.). Online dispute resolution. Retrieved from https://cedr.modria.com/