In our current age of technology, it seems like everything is going at the speed of light. AI is turning into an arms race between the most powerful governments in the world. Multi-billion dollar acquistions are being made every other week for bigger data centers and more GPUs. AI-powered tooth brushes (yes, really). And yet, there are still industry moving inversely at the speed of light, the most notable of which being grocery retailers. While everyone else is using AI to optimise every activity in their value chain, grocery stores are using the same systems from the 1980s, and relying on processes that were only fit for the early 2010s. Amongst the most neglected being in-store marketing.
Despite living in a time where advertising and marketing is more important than ever, grocery stores still rely on traditional marketing processess. Although there were some improvements made with digital signage, by the time a campaign is produced, the consumer behaviour, seasonal trend and inventory that was informing the creation of the campaign would eventually become irrelevant. This inefficiency leads not only to loss of potential sales, but food waste from spoiled food.
ShelfSense was created to tackle this issue. ShelfSense is an AI-powered ad campaign generation tool that helps grocery retailers reduce marketing costs, increase unrealized sales and prevent food waste. It monitors real-time data such as weather, season, time, temperature, inventory, consumer trends and more to inform our fine-tuned model to generate an ad campaign for the grocery store. Every month, our model is returns to the fine-tune process to create a feedback loop where the model can learn from its own campaigns and assess which ones worked and which ones did not.
ShelfSense allows grocery store retailers to reap the biggest benefit of AI-integration: automation and optimization. Currently, marketing teams spend days or weeks to bring a campaign from ideation to publication. In the middle of this process, they have to coordinate with multiple stakeholder such as management, IT departments, and third party creative agencies. Even then, by the time the ad is live, it become irrelvant due to the ever changing market environment. With ShelfSense, the ad creation cycle reduces from days/weeks to minutes, leaving marketers extra time to focus on more important activities like strategic planning. Additionally, ShelfSense creates campaigns that is real-time and data driven. It will generate campaigns that are optimised for conversion considering the relevant market data and inventory. Finally, it can prioritise promotion of soon-to-be expired inventory goods, which addresses the key sustainability issue of food waste.
Since the industrial revolution, grocery stores have been the backbone of societies. An abundance of food all in one place is something we take for granted. In an ever changing world, it is often these essential brick-and-mortar stores that tend to lag behind, but with ShelfSense, grocery stores can also enjoy the benefits of AI, one item at a time.
Students: Rajan Sapkota (788338rs), Simon Skarka (657510ss), Floris van Zeijl (585810fz)
PolicyPal
17
October
2025
5/5 (1)
Building PolicyPal: what we made, why it matters (and what is next)
If you have ever tried to read GDPR on a Friday afternoon…. We kept hearing the same story from small businesses: we are drowning in rules, we do not have a lawyer, and Google is not cutting it. That is the seed of PolicyPal, a lightweight, GenAI-powered helper that turns legalese into plain, cited answers you can actually use.
The problem
We mapped the SME reality: they are 99% of EU businesses, yet they carry a huge chunk of compliance work. Most teams either outsource (expensive), copy templates (risky), or just… wait out consequences. Add to that a flood of new EU regulations, GDPR, CSRD, the AI Act, each written in dense legal language and updated regularly. For small companies without in-house lawyers, figuring out what actually applies to them can take hours, sometimes days. The result? Missed deadlines, mounting costs, and a constant low-level anxiety about getting something wrong.
Some facts we found out
99% of all EU businesses are SMEs — employing over two-thirds of the workforce.
SMEs carry ~90% of total EU administrative compliance costs (≈ €200 billion/year).
55% of SMEs say regulation is their biggest barrier to growth.
43% have delayed expansion or digitalisation because of legal complexity.
The global RegTech market is expected to grow from $16 billion (2024) to over $33 billion (2029).
49% of surveyed SMEs already use technology for 11 or more compliance activities, showing both awareness and room for smarter tools.
What we built (so far)
Our prototype is a simple chat-style Q&A. You ask normal questions (“Do we need a DPO?”), and PolicyPal pulls from official Dutch GDPR texts, then drafts a short answer that makes complicated questions simple. Under the hood, we use retrieval-augmented generation (RAG): retrieve first, generate second, show your work. It is not wired to a live backend yet, but the experience, from question to cited answer to quick export, is there.
Who it is for
We are starting with Dutch SMEs, SaaS/IT, e-commerce, agencies, accountancies, and small clinics. The primary users are IT and HR managers who need fast, paste-ready answers with sources for tickets, policies, and stakeholder updates. Freemium lets teams try basic answers; upgrades unlock things like memo exports and audit trails.
What surprised us Even without direct SME testing, a few things stood out while building and discussing the concept. First, how huge the gap still is between regulatory language and what small business owners actually understand it is wider than we expected. Second, the number of existing “compliance tools” that claim to simplify things but end up just moving the complexity somewhere else. And third, how many SMEs openly admit they rely on guesswork or outdated templates to stay “compliant enough.” It confirmed our hunch that the real problem is not access to information, it is access to clarity.
The rough edges
It is a front-end demo for now. Risks are designing around: hallucinations, privacy, and over-reliance. We are aligning with ISO/IEC 27001 practices and the NIST AI RMF. For edge cases, there is the possibility of adding an optional human review layer to the business; however, that is not the main scope.
What we learned Working on PolicyPal showed us that solving a regulatory problem is not just about adding AI; it is about understanding how people experience complexity. Our research made clear that SMEs do not suffer from a lack of information, but from a lack of structure and confidence in using it. Regulations are written for legal professionals, not small business owners, and technology alone cannot fix that gap without thoughtful design.
We also learned that transparency matters as much as accuracy. Features like citations, disclaimers, and visible sources are not just “nice to have”; they determine whether users trust the answer at all. In that sense, PolicyPal became less about automation and more about building digital trust in an area where mistakes are costly.
Finally, this project helped us understand how business models and technology choices shape each other. RAG-based systems, freemium adoption, and compliance verification are not just technical or commercial decisions; they reflect how a tool positions itself between accessibility and accountability. For us, that intersection of design, trust, and value creation is the most important insight we are taking from the work.
Bring style and sustainability together with Salon
17
October
2025
No ratings yet.
Most people have gone through the moment of staring at a closet full of clothes yet feeling like they have no outfits to wear. Individuals in developed countries own many pieces of clothing, partially due to fast-fashion social media trends, which causes great harm to the environment. These problems of the overwhelming choice of possible outfits to create from your garments and many aspects of unsustainability in the fashion industry are both addressed by Salon, an AI personal stylist platform which promotes circular fashion.
The problem: environmental harm and an overwhelming number of choices to make
The number of times pieces of clothing are worn has decreased by 35% over the last fifteen years (Nizzoli, 2022). On top of that, people buy 60% more clothes than in the year 2000 (Igini, 2024). As a result of this, individuals throw away quite a few of their clothes every year, which mostly end up in landfills and lead to environmental pollution (KEMI Swedish Chemicals Agency, 2014). At the same time, owning many garments creates an overwhelming amount of outfit choices (Schwartz, 2015).
The solution: Salon
To help mitigate this global problem, Salon’s mission is to increase wear per garment and discourage unnecessary purchases in a unique way by changing the behaviour of individuals. By using Salon, users get a digital version of their wardrobe, an AI-powered personal stylist, virtual try-ons, a fashion community, and a (second-hand) marketplace all in one platform. With these features, the platform addresses both the fast-fashion crisis and the daily challenge of selecting an outfit through an AI-powered fashion ecosystem that transforms the way users engage with their existing wardrobe and promotes sustainable consumption. This approach is in line with the United Nation’s Sustainable Development Goal 12 of Responsible Consumption and Production. Salon specifically focuses on responsible consumption by aiming to reduce waste by extending the lifespan of clothes through smart styling and increased reuse of garments. This will be done by shifting consumer behavior by encouraging people to be more resourceful with their existing wardrobe, instead of frequently purchasing new clothes.
Market fit
Salon operates at the intersection of fashion technology and environmental sustainability. These markets were valued in 2024 at 239 billion USD and 10 billion USD respectively, at compound annual growth rates of 6.3% and 9.5%. Its primary target demographic is Generation Z because they are relatively seen as the most active in these markets. They are “technologically native” (Mehta, 2022), fashion-focused, and environmentally aware. However, they experience an attitude-behaviour gap between their vision on sustainability and their actual fashion purchasing habits (Prashar & Kaushal, 2025). Salon bridges this gap by merging environmental sustainability and consumer convenience with its AI-driven services. This intersection of these markets highlights a significant societal and commercial opportunity.
Conclusion
Salon demonstrates how GenAI can transform fashion consumption from frequent purchasing and waste of clothes to mindful styling with sustainability in mind. By turning wardrobe management into a digital ecosystem built on the principles of sustainable consumption and user convenience.
References
Igini, M. (2024, May 30). 10 concerning fast fashion waste Statistics. Earth.Org.
KEMI Swedish Chemicals Agency. (2014). Chemicals in textiles – risks to human health and
the environment. In KEMI Swedish Chemicals Agency (No. 6/14). https://www.kemi.se/en/publications/reports/2014/report-6-14-chemicals-in-textiles
Prashar, A., & Kaushal, L. A. (2025). Nudging sustainable fashion choices: An experimental
investigation on generation Z fashion consumers. Acta Psychologica, 253. https://doi.org/10.1016/j.actpsy.2025.104727
Schwartz, B. (2015). The Paradox of Choice. In Positive Psychology in Practice Part II:
Values and Choices in Pursuit of the Good Life (pp. 121–138).
ComplyAI: The Solution for Regulatory Compliance
17
October
2025
No ratings yet.
Nowadays, the number of new laws and regulations is exploding. Last year over 60,000 regulatory updates were recorded for companies to comply with. This makes compliance both expensive and time-consuming for companies, with compliance costs rising by more than 50% in the past decade. Many companies struggle to keep up and risk large fines when failing to comply. Our team addressed this challenge by developing ComplyAI, a generative AI-powered compliance assistant that helps organizations stay compliant more efficiently for companies across all industries and preparing their reports in the right format.
ComplyAI changes the way organizations could handle compliance. Instead of reading long legal texts and manually updating policies, the system scans regulatory databases, summarizes changes and checks which ones apply to each company. It then produces action lists, templates and deadlines for these companies while flagging any areas that still need human review. This “human-in-the-loop” approach keeps experts in control while saving time and reducing errors and maintaining human accountability.
ComplyAI would operate in the fast-growing Regulatory Technology sector, The sector is expected to expand from $15 billion in 2024 to $83 billion by 2032. The ComplayAI-tool would be especially valuable for medium and large firms in heavily regulated sectors like finance, healthcare, and energy. In these industries compliance is complex and mistakes are costly resulting in a lot of room for efficiency improvements.
Our analysis shows that ComplyAI can cut manual compliance work by 50–65% and lower total costs for companies by 30–40%. Continuous monitoring compliance also helps prevent violations which potentially reduces the risk of major fines. At the same time, employees can focus on meaningful tasks such as interpretation and strategy, which improves job satisfaction and overall productivity in the workplace.
Our prototype, which is available at regulatr-ally.lovable.app, demonstrates how employees can upload documents, get instant compliance summaries and generate actionable reports. Our platform shows how AI can be responsibly integrated into corporate workflows to support professionals rather than replace them.
In short, ComplyAI helps organizations stay compliant faster, cheaper, and smarter. It turns the challenge of regulatory change into an opportunity by helping companies build trust, stay competitive, and focus on what really matters.
The Smartest Colleague in the Room: AI as the New Internal Expert
17
October
2025
No ratings yet.
In a reality where technology is almost an intrinsic characteristic of our society it would seem like our communication channels are super efficient, right? Well… it is not fully wrong, yet it is still far from perfect efficiency, especially in the business world. Corporations and workforce are bigger than ever but somehow people feel more disconnected from work than before. The numbers are shocking as almost 80% of workers feel disengaged from work activities, thus creating a negative impact on workspaces.
Despite our different work experiences, we found common ground on a challenge that we all faced: not having internal company knowledge. Have you ever had questions that you did not know how to answer? “Who can I contact for IT support? Who do I contact for this matter?” Hours pass by as you reach out to different colleagues, before you obtain the information you need. These questions may seem like minor inconveniences, but across an entire organization, these inefficiencies compound into significant productivity and time losses.
What already feels like a burden is augmented even more once you reach this phase and a big catalyst for this is miscommunication. Big companies are successful but they can also be their biggest challenge if they ignore foundational problems. Luckily, we were born on the same timeline as our new digital friend and saviour of proper communication channels, the AI-Agent.
One significant takeaway from this project was the potential of a well-designed AI assistant to revolutionise knowledge management and internal communication within organisations. By integrating structured company data and deploying an intelligent, conversational interface, employees can access information up to 30% faster, significantly improving productivity and onboarding experiences. Moreover, aligning the AI implementation with clear SMART goals (Doran, 1981) ensures measurable impact and organisational value. It gives employees metrics to look for and creates a standard for further improvement.
When it comes to the creation of our business model, creating a user-centric design that meets regulatory compliance offers a sustainable advantage. It increases productivity and trust within the organisations, especially for those operating in highly regulated sectors such as financial services.
Let’s talk about impact!
The numbers are loud and clear. Our AI agent solution transforms operational inefficiency into measurable productivity. Firstly, it cuts the time agents spend searching for information by 30% which is a big chunk of time saved alone (Bula et al. 2025). Secondly, onboarding time is shortened by 25% as agents are no longer reliant on colleagues alone for asking their questions (Bula et al. 2025). Furthermore, to ensure frequent quality and compliance updates clever use is made of self-enforcing feedback loops as discussed by Jullien et al. (2021). These feedback loops ensure updates and continuous learning.
The AI agent provides a higher quality customer service desk. In a world where time is the most valuable resource, our AI agent will give it back to you.
References:
Bula, A., Torres, B., Hu, M., & Fennis, W. (2025). Innovating business models with Gen-AI [Unpublished academic report]. MSc Business Information Management, Erasmus University Rotterdam.
Jullien, B., Pavan, A., & Rysman, M. (2021, July). Two-sided markets, pricing, and network effects (TSE Working Paper No. 21-1238). Toulouse School of Economics.
Big Marketing Power for Small Businesses: Introducing Marvia in the Omnicom Group
17
October
2025
5/5 (1)
An important part of running a company is to have a high brand awareness across the firm’s target audience. Brand awareness is created through extensive marketing efforts. Marketing activities such as customer engagement, market research, and marketing campaigns can contribute to a long-term competitive advantage if carried out effectively and efficiently (Bocconcelli et al., 2018). Yet, most small and medium-sized enterprises lack a clear marketing strategy because of financial limitations, lack of knowledge, and time constraints. This is why we created the new virtual marketing assistant, Marvia.
Marvia can help SMEs structure their marketing activities by creating marketing strategies, posts, promotional texts, marketing schedules, budgets, and more. This Software-as-a-Service platform allows enterprises to overcome the barriers they experience when it comes to creating a well-thought-out marketing strategy by putting all the skills and knowledge required in a single AI tool.
The tool will have a chat function, similar to popular generative AI tools such as ChatGPT or Google Gemini. Here, users can use prompts to communicate their type of business, target customer groups, marketing budget, and wishes and demands to the AI tool. Marvia will inspect the demands of the SME and will create a marketing plan based on all the needs of the company. It will also help by creating posts, thinking of potential captions or stories, and supporting the SME with all their marketing questions and ideas.
The marketing assistant tool will be part of the larger marketing company, Omnicom Group. By anonymizing these plans, Marvia can be trained on data produced by Omnicom. This means that there is ample training data available by using old marketing plans that Omnicom has produced. By only training the model on in-house data, the branding of Omnicom will show in all the work created by Marvia. Rigorous training processes, such as contextual embedded learning, instruction learning, convolutional neural networks, and diffusion models, will ensure the quality of Marvia’s work will be as close to the real-human product as possible.
To keep up with the success of their marketing campaign, Marvia has a campaign dashboard. This dashboard will show all important financial metrics, such as budget spent and funds left, as well as important marketing metrics, such as reach, interactions, and conversion rate. The dashboard will show the next planned marketing activities to remind SMEs of their upcoming posts and events.
By creating an easy-to-use and intuitive marketing tool, SMEs will be able to step up their marketing game. The knowledge gap between large companies and SMEs will shrink, since the smaller firms will have 24/7 access to a marketing expert who can help them create the perfect marketing campaign to boost their brand awareness, reach, and ultimately, profit.
Team 04
Guus de Bruijn (604863gb)
Mischa Terlaan (764007mt)
Paolo Cozzolino (579063pc)
Sina Bergjürgen (764860sb)
References:
Bocconcelli, R., Cioppi, M., Fortezza, F., Francioni, B., Pagano, A., Savelli, E. and Splendiani, S. (2018). SMEs and Marketing: A Systematic Literature Review. International Journal of Management Reviews, 20: 227-254. https://doi-org.eur.idm.oclc.org/10.1111/ijmr.12128
DuoMate: Conversational Fluency with GenAI
17
October
2025
No ratings yet.
Duolingo is one of the world’s most influential and leading language learning apps with over 100 million users (Guerrero, 2024). However, it comes with one key gap: conversational fluency. Our team identified this “speaking gap” as a limitation that can threaten Duolingo’s value proposition. We propose the integration of a GenAI conversational partner called DuoMate as a solution that can strengthen Duolingo’s competitive position.
DuoMate’s LLM model will be trained using Duolingo’s large proprietary base (Zacks, 2025) to ensure optimal delivery that is personalized, accurate, contextually relevant, and adaptive. It also integrates speech-to-text and text-to-speech in different real-life contexts to simulate natural conversations. Its instantaneous feed-back on grammar, vocabulary, and pronounciation aims to elevate users’ confidence with speaking a foreign language. Together, the model provides an adaptive feedback loop that bridges the gap between learned lessons and practical implication.
Currently, Duolingo operates on a freemium model. We suggest embedding the DuoMate feature with Duolingo’s recently introduced subscription, Duolingo Max (Mlot, 2023), to increase its perceived value and encourage free users to switch to a premium subscription model. By attracting both individual learners and corporate collaborations, this business decision enhances Duolingo’s revenue potential. We estimate a 41% revenue growth in the first year post-release of this feature. Duolingo’s paid subscribers are projected to grow from 20.3 to 28.3 million. With an expected window of 6 month development period prior to the launch, the estimated total development costs and first-year costs come up to around 94.7 million USD.
Upon conducting market research, we have found that the global online language learning market has been rapidly growing and is projected to reach a valuation of 91.6 billion USD by 2030 (Markets, 2024). Furthermore, there is a growing shortage of qualified human instructors – 44 million short by 2030 (Taylor, 2025) – that places a strain on language course pricing and fixed schedules. Duolingo has the opportunity to minimize the shocks of this shortage by leveraging GenAI to deliver scalable, highly personalized lessons at a lower cost and much greater flexibility. We expect early adopters to range from intermediate and advanced learners to corporate clients.
By addressing Duolingo’s critical learning gap, it can transform its platform’s value proposition from a “fun, gamified learning app” to a “conversational fluency buddy” becoming a complete language learning ecosystem. As part of Duolingo’s existing AI-first strategy which led to a 51% surge in daily active users contributing to its 1 billion USD revenue forecast for 2025 (Daigle, 2025), DuoMate plays a pivotal role in Duolingo’s evolution and growth. Thus, DuoMate serves as an innovative growth catalyst that benefits both Duolingo and its users by offering high quality, tailored lessons to each individual user while expanding Duolingo’s revenue potential, strengthening user loyalty, and reinforcing its dominance in the AI-powered education market.
Guerrero, N. (2024, October 5). Good, free, fun: The simple formula that has made Duolingo a daily habit for millions. https://www.bbc.com/worklife/article/20241004-the-simple-formula-that-made-duolingo-a-daily-habit-for-millions
Mlot, S. (2023, March 17). Duolingo’s MaX subscription uses GPT-4 for AI-Powered Language learning. PCMAG. https://www.pcmag.com/news/duolingos-max-subscription-uses-gpt-4-for-ai-powered-language-learning#:~:text=Duolingo%20jumped%20on%20the%20generative,conversations%20with%20in%2Dapp%20characters.&text=The%20Explain%20My%20Answer%20feature,to%20go%20for%20a%20hike.%22
Marriott – Personalizing Hospitality with Generative AI
17
October
2025
5/5 (1)
Marriott International has long led the hotel industry with its asset-light business model, global brand consistency, and impeccable service. Making it one of the biggest hotel chains in the world. However, with increasing competition from platforms such as Airbnb and the shift into an intensified experience economy, guests’ demands are shifting. Therefore, to keep up with the competition and remain a market leader, Marriott has to innovate.
Our project presents Marriott with this solution, namely, implementing Gen-AI in the pre-stay booking phase with the Gen-AI pre-stay travel planner and during the stay with the in-stay travel assistant.
Here is how it works: The pre-stay travel planner allows guests to get a tailored overview of activities and plans at their intended stay, which is specific to their customer profile and trip details. Think of it as a curated co-pilot that kicks in the moment you book. It builds a tailored itinerary based on your profile, destination, dates, budget, and preferences. It then pulls context like weather and local events, and you can chat with it to refine plans. Resulting in a seamless trip booking experience and significantly reducing the hassle of planning it yourself.
Then, during the stay, the Gen-AI in-stay assistant integrated with the larger Marriott service system allows guests to get immediate customized answers and recommendations tailored to the stay. When questions get too complex, it recognizes this and escalates to a Marriott employee, so you get speed without losing Marriott’s trademark human touch.
While these improvements increase the value customers receive from booking with Marriott, these implementations simultaneously provide Marriott with several key benefits.
It enables Marriott to profit from new monetization and revenue streams. It does this by automatically up-selling add-ons, cross-selling, and finally engaging in Gen-AI partnership deals based on commission. Making it a truly lucrative implementation.
It increases operational efficiency: Offloading routine queries to AI, which frees staff to focus on high-impact service moments.
And finally, it increases customer loyalty and raises switching costs. As guests keep booking, MarriottAI learns preferences and habits, making future stays feel increasingly personalized. In return, leading to more engagement with its loyalty program, Bonvoy.
To successfully roll out this GenAI integration in such a large brand, the implementation should be phased in 3 key stages:
Months 0–12: Will focus on a Pilot in ~10 diverse hotels, where key adoption metrics are tracked, and recommendations are refined.
Months 13–24: Expand the pilot to 100–150 properties, deepen system integrations, start marketing to Bonvoy members, and keep training on real usage data.
Months 25+: Roll out globally with multilingual, location-specific tuning
In summary, by using AI strategically and augmenting the stay with a smart layer that plans with you and adapts to you, Marriott is enhancing its ability to do what it does best. Guests get trips that feel curated, teams get time back for the moments that matter, and Marriott turns every booking into a relationship that gets better with each stay, allowing it to lead in an experience-first market.
Team 10
From Weeks to Days: How GenAI Can Transform Traditional Banking at ING
17
October
2025
No ratings yet.
Waiting three to five weeks for a loan approval feels antiquated in an era of instant digital services. Yet this is the reality for ING customers, a striking contradiction for a bank that pioneered digital banking in Europe (ING Groep N.V., 2024). With a €164.3 billion lending portfolio and advanced digital infrastructure, ING provides an ideal case for showing how Generative AI can resolve this paradox and help traditional banks defend against fintech disruption.
The ING Innovation Paradox
ING serves 8 million customers, 84% of whom prefer mobile banking, and invests €324 million annually in software (ING Groep N.V., 2024). Yet its loan approvals remain constrained by manual processes consuming 8–10 hours per application. Traditional workflows still require three to five weeks for decisions (McKinsey & Company, 2018).
Meanwhile, fintech competitors leverage automation for near-instant approvals. In the Netherlands, fintech lending doubled to €2.3 billion in just two years (De Nederlandsche Bank, 2024), threatening ING’s market position as customers who experience poor digital service rarely return (Accenture, 2025).
The GenAI Solution
The proposed solution applies Generative AI to analyze unstructured financial documents while maintaining explainability and regulatory compliance. Using natural language processing (NLP), GenAI interprets payslips, invoices, and income records to generate standardized, transparent risk assessments (Al-Hchemi, 2024).
Figure: Prototype of self-developed ING GenAI Loan Assistant
The three-stage process includes secure data ingestion, AI-driven document classification, and generative summarization that produces explainable risk profiles. Critically, humans remain in control, AI accelerates analysis without replacing judgment, ensuring full compliance with GDPR and the EU AI Act (European Parliament, 2023).
Transformative Impact
Processing time drops by 85%, from several weeks to just three days. Per-loan costs decrease 78%, from €450 to €100, as automation reduces manual work from eight hours to fifteen minutes (McKinsey & Company, 2023; Deloitte, 2019). Over three years, this transformation generates an estimated €700 million in cumulative revenue and €10.5–12.6 million in annual cost savings.
Beyond efficiency, AI-driven credit scoring enables 15–20% more approvals by fairly assessing self-employed individuals and those with limited credit histories (Fares et al., 2023). This expansion enhances financial inclusion while profitably serving previously underserved segments.
The model also demonstrates the economics of information goods (Shapiro & Varian, 1998): an upfront investment of €18–23 million yields dramatically lower marginal costs over time, creating sustainable competitive advantage through operational learning and scale.
Strategic Defense
A Business Model Canvas analysis (Osterwalder & Pigneur, 2010) reveals deep transformation across ING’s value structure, from automated key activities and AI partnerships to improved customer relationships, expanded target segments, and scalable cost economics.
Implementation challenges remain. Successful adoption requires robust change management (Gutzwiller et al., 2024), enhanced cybersecurity (FINMA, 2023), continuous bias audits, and strict adherence to evolving regulatory standards (Bank for International Settlements, 2021). Yet these challenges are outweighed by the strategic imperative to lead the AI transformation rather than react to it.
Conclusion
Integrating Generative AI is both technically feasible and strategically essential, fully aligned with ING’s Think Forward strategy. By embedding AI while retaining human oversight (Vial, 2019), ING can transform a core operational weakness into a durable competitive advantage.
The question is no longer whether AI will transform lending, it already is, but whether traditional banks will lead the transformation or be disrupted by it. ING possesses both the capability and the strategic imperative to choose leadership.
Team 22: Elias Stad 715336, Helen Hu 537226, Mihai Tulbure 765023, Vinita Ohm 787879
—
References
Accenture. (2025). Global banking consumer study 2025: Banking advocacy powering growth. https://www.accenture.com/us-en/insights/banking/consumer-study-banking-advocacy-powering-growth
Al-Hchemi, L. H. (2024). Evaluating Generative AI in enhancing banking services efficiency. Економічний Форум, 47–54. https://doi.org/10.62763/ef/4.2024.47
Bank for International Settlements. (2021). Artificial intelligence and machine learning in financial services. Basel Committee on Banking Supervision. https://www.bis.org/bcbs/publ/d537.pdf
De Nederlandsche Bank. (2024). Annual report 2024. https://www.dnb.nl/en/publications/publications-dnb/annual-report/annual-report-2024/
Deloitte. (2019). AI leaders in financial services: Common traits of frontrunners in the artificial intelligence race. https://www2.deloitte.com/us/en/insights/industry/financial-services/artificial-intelligence-ai-financial-services-frontrunners.html
European Parliament. (2023). EU AI Act: First regulation on artificial intelligence. https://www.europarl.europa.eu/topics/en/article/20230601STO93804/eu-ai-act-first-regulation-on-artificial-intelligence
Fares, O. H., Butt, I., & Lee, S. H. M. (2023). Utilization of artificial intelligence in the banking sector: A systematic literature review. Journal of Financial Services Marketing, 28(4), 835–852. https://doi.org/10.1057/s41264-022-00176-7
Gutzwiller, E., Moutiq, N., & Pearce, M. (2024, June 14). Ensuring a successful transformation through non-IT change management. Deloitte. https://www.deloitte.com/ch/en/Industries/financial-services/blogs/navigating-tech-enabled-transformation-of-core-banking-processes-part-3.html
ING Groep N.V. (2024). Annual report 2024. https://www.ing.com/Investors/Investors/Financial-performance/Annual-reports.htm
McKinsey & Company. (2018). The lending revolution: How digital credit is changing banks from the inside. https://www.mckinsey.com/capabilities/risk-and-resilience/our-insights/the-lending-revolution-how-digital-credit-is-changing-banks-from-the-inside
McKinsey & Company. (2023). The state of AI in 2023: Generative AI’s breakout year. https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai-in-2023-generative-ais-breakout-year
Osterwalder, A., & Pigneur, Y. (2010). Business model generation: A handbook for visionaries, game changers, and challengers. John Wiley & Sons.
Shapiro, C., & Varian, H. R. (1998). Information rules: A strategic guide to the network economy. Harvard Business School Press.
Vial, G. (2019). Understanding digital transformation: A review and a research agenda. The Journal of Strategic Information Systems, 28(2), 118–144. https://doi.org/10.1016/j.jsis.2019.01.003
Personalization with Purpose: Zalando’s Digital Stylist and the Future of Conscious Shopping
17
October
2025
No ratings yet.
Team 44: Aysenaz Cimsit (633875); Marianna Tabaczyk (647650); Sasha Mballa Ambani Ledji (777624); Sem Gelderblom (647994)
Online shopping marketplaces are responsible for the unsustainable practices of fashion consumption that contribute to the deterioration of environmental conditions globally. As a fashion marketplace, Zalando has the opportunity to reduce over consumption by leveraging a virtual try-on (VTO) tool to reduce item returns. Zalando itself indicates that about 50% of ordered items are returned (Zalando, 2022). Flexible return policies also encourage buyers to shop more, and accelerate supply chain activities that worsen environmental conditions (Velazquez & Chankov, 2019). It can thus be explored how Zalando can be positioned as a leader in the sustainable consumption of fashion. The VTO tool, Digital Stylist, creates a personalised experience for buyers to visualise digitally how their chosen items would look on them before purchase. The gap of doubt is therefore minimised, and subsequently return rates are reduced and customer satisfaction is improved. The pilot project with Zalando aims to promote an IT-driven change that implements AI and machine learning to transform the impersonal buying experience of online shoppers. In effect, Zalando management gained insights on buying behaviour and customer satisfaction to improve their competitiveness in the market.
Current VTO tools are generic and use an avatar to model items on a body. The Digital Stylist prototype initially prompts users to upload a full body image, their body measurements, and their selected items. Using language models and generative models, the tool produces a render of the user wearing their selected items. An AI assistant additionally serves as an online personal stylist to support the interactivity of the tool. Two versions of the tool are created to serve audiences with a significant difference in digital literacy and shopping habits. The AR-ready version gives customers the ability to view how clothing behaves and fits from multiple angles. The light version is optimised for speed by providing lower quality outfit previews directly on uploaded 2D photos of the individual. At the pilot testing phase of the launch, the prototype will be limited in functionality and capacity to allow for multiple testing iterations. The next phase of the product launch focuses on the tool at scale, which requires an in-house development of the tool to increase its capacity and processing power. The two phases are guided by the following key metrics; an 8% decrease in return rates, 15% increase in conversion rates, and a 10 point increase in Zalando’s Net Promoter Score. The metrics not only evaluate the economic effect, but also position Zalando as a leader for responsible consumption and production in the fashion industry.
The success of the launch requires management to effectively create alignment between teams to manage the potential areas of risk; security of personal data, public backlash on image alteration, and the complexity of the tool in comparison to competitors. Management must require research to be conducted on the potential risks, and for alignment to exist between the required elements of the product and customer needs.
The Digital stylist will not only allow Zalando to strengthen their competitive position through innovation, it will also help the world as a whole by contributing to lowering world wide pollution on a planet that is in desperate need of help.
References:
Velazquez, R., & Chankov, S. M. (2019, December 1). Environmental Impact of Last Mile Deliveries and Returns in Fashion E-Commerce: A Cross-Case Analysis of Six Retailers. IEEE Xplore. https://doi.org/10.1109/IEEM44572.2019.8978705
