PROBLEM
Foundation Models are emerging as a new paradigm in AI research &
commercialisation. While this opens up possibilities for radically innovative
solutions and significant value creation, startups are challenged with finding
unique & differentiated ways to leverage the technology, while simultaneously
mitigating potentially negative consequences.
Despite the rising prevalence of machine learning (ML) in products, designers
face challenges in creating solutions that make the best use of the new
possibilities. Designers building AI products are rarely involved in problem
setting & value finding, primarily solving human-AI interaction problems.
Additionally, common human-centered, customer feedback based innovation
approaches hinder radical innovation.
Lastly, AI Ethics and responsible innovation continues to be an afterthought.
Product designers are seldom involved in mitigating the potentially negative
consequences of such products.

APPROACH
This thesis combines literature study with a "research through design"
approach to explore ways to address these problems. Working in collaboration
with a startup trying to leverage Large Language Models (LLMs) in their
products,I use empirical research to engage in the process of finding radically
innovative opportunities for using LLMs to create value for customers.
I study risk assessment practices to explore how designers can anticipate
risks in the discovery phase of product design. I then design & test tools that
can support other designers in the future.

OUTCOMES
This thesis makes multiple contributions to further the research on
technological & responsible innovation. It documents my process for finding &
analysing user insights to support the discovery of potentially radical
innovations. Additionally, I explore the process of finding value propositions
that leverage Foundation Models, and their potential risks, early in the design
process. The thesis also records how I design a process for discovering and
anticipating potential risks of harm, and how I developed a pair of canvases
and card decks to support future designers.
The thesis supports product designers in repeating these processes through
the pair of canvases and card decks. These help them to collaborate with
engineers, and contribute to the innovation & risk mitigation processes more
effectively. The discussions focus on observations & recommendations that
can further aid them.
To help designers innovate responsibly, this thesis brings the discovery of
opportunities and risks of using Foundation Models into the same
conversation. The designed deliverables and processes showcase how both
aspects of technology innovation can be tackled in similar ways.
Through all of the above, this thesis showcases the relevance of designerly
ways of thinking and doing to the fields of radical innovation, risk
management, and Foundation Model based product development. ... Read more

This report describes the results of a graduation project, which was completed at Hombli, a brand of Avanca International B.V. Hombli offers smart home products and has the ambition to become a leader in this domain. In order to achieve that the company wants to create Intellectual Property in the future and have a solid brand. Hombli has two issues however First of all Hombli’s current positioning risks losing its uniqueness. Competitors also claim to be simple to use, easy to install and safe. Secondly, the way it currently creates new products is sensitive to being copied by competitors. Hombli imports products from China, adjusts them to their liking and integrates it into their ecosystem of devices. Many of these smart home products are also available to competitors, making it hard¬er for Hombli to stand out. A question rises when looking at these obstacles: How can Hombli become a household name and create I.P.? To answer this question a two step approach is proposed: First a future direction will be designed, resulting in a clear and differentiating brand story for Hombli. Second a new way innovation process will be designed as well as improving the existing way of innovating. By doing trend research and using the trend spotting method (Simonse, 2017), a future scenario is created. Summarized it resulted in the following: Because of an increase of people suffering from mental stress and the home becoming an even more prominent space in people’s lives, it makes sense for Hombli to facilitate well being at home. Then the way the company currently innovates is analysed. The existing process which focusses on sourcing products has some disadvantages. Lack of communication, no prioritisation and confusion about responsibilities result in unnecessary slack in the process and in turn hinders innovation. This then also leaves even less room for any exploratory activities. Principiles from the User Centered Design method from Abras, Maloney-Krichmar & Preece (2004) are used to redesign the current innovation process together with the employees that actually use it. During the weekly sessions the different stages of the process are discussed, altered and put to practice with actual products. The process is altered to become more like the Stage Gate innovation model from Cooper (2008). Advice is given on how to implement this in the organisation. Then the process focusing on discovering new ideas is introduced. Throughout the steps assumptions regarding the product direction are validated. This is done through initially desk research, then interviews with target users and an individual co-creation session. The product idea that results from that session is put to the test using a Smoke Test (Cops, 2020). A method used often by start-ups to see if their value proposition gains traction. The results of this step are positive and give a green light for next iterations. Then recommendations are given to further expand the proposed changes in the company to have a lasting positive effect in the future. Concluding, limitations of this research are also shared. ... Read more

This project is a collaborative effort between graduating student Johan van der Schaaf from Delft University of Technology, BoP Innovation Center and client organization Mhogo Foods. The project is conducted mainly in Nairobi, Kenya and focuses on building a marketing pilot for Mhogo Foods to increase access to nutrition in low-income markets. The project taps into the challenge of marketing nutrition and the difficulty of measuring effectiveness of marketing campaigns. After a short literature study and 15 interviews with the target audience that provided several insights, a design-led approach was used to create a tag line that resonates with the target audience. The combined insights resulted in a tag line that formed the key message of the campaign: ‘Mlo Wetu Halisi’ (in English: ‘Eating, in our traditional way’). The design-led approach turned out to be helpful in targeting the message to the preferences of the target audience and ensuring the tag line is inclusive. During implementation of the campaign, all materials and interventions were based on the predefined key message. The implementation of the campaign consisted of a mix of proven offline activities to gain the first market traction with digital experiments to explore new marketing opportunities. The digital experiments consisted of streaming recipe videos through Wi-Fi platforms on local buses and distributing digital coupons through several digital channels that could be redeemed at local shops. During the campaign, different digital channels were tested through the digital coupon program. An initial setup, where coupons would be distributed through Facebook, Whatsapp and the Wi-Fi platforms on buses highlighted the challenge of reaching customers through the digital channels while also educating them on the program. A second iteration, consisting of a hybrid between offline and online channels, resulted in more tangible data, with customers showing understanding and interest in the coupon program. An additional key insight was that the Wi-Fi platforms turned out to be the only digital channel that could both reach the target audience and keep the attention span of consumers for a prolonged period of time. Finally, a tool is presented that can be used in future marketing pilots to assist in building the campaign and in designing strategies for measuring effectiveness of digital channels. ... Read more

The construction and logistics sectors in the Netherlands are rapidly increasing in size, but unfortunately are also part of the most hazardous industries. One part of safety which needs to be addressed is manual handling related incidents. The moving of objects around a worksite through utilization of an employee’s body increases the exposure of the employees to hazards, especially if done incorrectly, and will increase costs for both employee and employer. This thesis explores the possibility of transforming the passive and reactive role of safety shoes into safety shoes that are capable of proactive (manual handling related) incident prevention, through utilization of smart technology. To gain insight into the problem, its context and the possibilities for the implementation of smart technology, an extensive literature review was conducted. In addition, practical field knowledge was gained through multiple series of semi-structured interviews and analysis of relevant cases. Research results show that the human factor plays a major role in the causation of occupational incidents and can be divided into three categories: the individual, the task and the organisation. As this human factor is either the leading cause or part of the cause for around 80% of all incidents, it is vital for incident prevention. Furthermore, current methods for detection and prevention (e.g. manual handling training and safety programs) have serious shortcomings which make them less effective tools for the reduction or elimination of manual handling related incidents. In addition to this, studies indicate the opportunity for (smart) technology to aid in overcoming these shortcomings. The above mentioned insights served as input for the synthesis of a concept design (smart safety shoes), which uses sensors and data analysis tools (e.g. machine learning) to identify and detect leading and lagging indicators for manual handling related incidents and subsequently is able to effectively communicate those insights to different parties: employees, employers, supervisors and training providers. The smart safety shoes can, in this manner, support current detection and prevention methods in their shortcomings. The before mentioned parties deploy the insights through a hybrid system: reactive incident prevention (improve the individual) and proactive prevention (improve task design and organization), which both, increase incident prevention. The concept design is accompanied by a roadmap outlining the general steps for the development of the concept. To conclude, smart safety shoes are the next step towards occupational incident prevention and potentially the first step towards smart, ubiquitous, occupational safety. However, further research is needed for the development of the smart safety shoes and the exploration of further possibilities. In addition to this, the principles behind the smart safety shoes could serve as a basis for further design research, to address other occupational safety issues and other industries. ... Read more

The aim of this thesis is to increase retention for Picnic by creating a habit-forming store. Currently, conversion rates are suboptimal. In areas where Picnic is active roughly half of all households downloads the app and registers. However, the conversion of this group to active customers is sub-optimal, meaning that not all users are retained. Sub-optimal retention rates hurt companies, as they miss opportunities for business growth, increasing their profitability and cost savings. This effects also holds true for Picnic where it takes three orders to break-even on that specific customer. Project context This thesis focusses on the Dutch market, as this is where Picnic as its strongest presence. In addition to that, the Dutch online grocery market is the most advanced in Europe. This thesis focusses mainly on the competition with physical supermarkets because the growth of the online groceries segment stems from consumers switching from physical supermarkets to online grocers. Increasing the retention rates has a positive impact on the users of Picnic and society as a whole: It increases the amount of free time for users, reduces food waste and emissions in inner cities. Theoretical foundation Retention Retention is driven by the perceived utility on one hand and switching costs on the other. Both these dimensions are influenced by the habits of users. Perceived utility increases as the users gain more experience with the product, leading to a faster and more pleasant experience. Switching costs are influenced by the investments a user makes in a product, and by uncertainty about whether other companies can provide similar value. Groceries Groceries are eminently a habitual consumption: They are high-frequency purchases Groceries have a limited variability over time Grocery shopping is a behaviour that is learned over a long period of time Therefore, the challenge of this thesis is to help more users successfully form a habit around using Picnic, thereby converting into active customers.

Habits
Habits are formed when both frequency and perceived utility are high enough. A habit has four elements: The trigger, action, reward and investment. Currently, most customers fail in the action element of the habit loop. This is due to either a lack of ability or a lack of motivation.
This theory on habits and its connection is visualized in the figure below.


User research
The theoretical framework was combined with extensive user research to define what type of users currently are able to form a Picnic habit.
The qualitative user research consisted of a combination of interviews, workshops, user testing sessions, surveys and concierge tests. The quantitative user research consisted of analyses of in-app user behaviour data, purchase data, demographic analysis and market size estimations.

Early market
Users that successfully form a Picnic habit, and thereby form Picnic’s early market have both high motivation and high ability to use the service.
The high motivation results from the users being time-constrained, having physical limitations, and experience going to the supermarket negatively. This group’s high ability is a result of their high ability to plan, their relatively predictable life, and the guidance in food they receive from recipes and diets. The majority of this group are families, which are also the most profitable customers for Picnic.

Product strategies
By combining the insights from user research with the theoretical framework, three product strategies were proposed that help more users form a strong Picnic habit.
First more different types of users are enabled to place orders by their increasing their motivation and ability. Once these users are enabled to order at Picnic, their frequency of interaction is increased. Finally, when these users are interacting with this app frequently, users remain engaged by an app that improves with each usage-cycle. This happens through both user- and Picnic- driven personalization.

Implementation of product strategies
These strategies must be embedded in the store team and Picnic organization in order to be effective. In order to do so, three organizational challenges must be overcome:
Challenges in aligning teams across the organization,
Challenges in creating buy-in for customer-focused projects
Challenges in autonomous decision making, due to large dependencies between teams.

Product design framework
To deal with these organization challenges, the strategies are translated into a product design framework. This framework provides the team with clear and inspirational guidance while being concise and measurable for the rest of the organization. This leads to companywide buy-in.
 
Validation
The framework was validated in two steps: In sessions with customers and internally. The results of these validation steps indicate that the framework is likely to be successful, but it can only truly be proven by implementing it and putting it to the test.

Implementation
To achieve successful implementation, the framework must be clearly communicated and embedded in the daily work of the team. The latter should happen in both creative activities and evaluative activities.

Conclusion
The framework helps the store team to set a course for a longer period of time. It helps the team focus on solving the right problems for its users, by providing a way to visualize these problems and the effect the solutions should have. The vision and strategies provide the team with direction and help stakeholders across the organization align. The framework is not a definitive ‘how-to guide’ on building a habit-forming store, and its implementation and the subsequent execution are crucial for its success.  ... Read more

This project focuses on the integration of User Research in a Qualitative approach into an Agile work environment by conducting an in-house Case Study with Werkspot (Home improvement platform in the Netherlands). Customer expectations have hit an all-time high globally (Salesforce Research 2018) and as such, companies are expected to provide experiences beyond products (physical or digital). This has impacted the way companies must operate going forward, by recognizing and addressing customer involvement throughout the development process. The challenge relies in integrating two seemingly contradicting processes Agile development (fast-paced) and Qualitative Research practices (slow-paced) into a seamless operation. The objective is to include the end-users early in the development process. In this way, Werkspot is able to increase the chances of success of product features by implementing a validation phase prior to development process. This project makes a research distinction into Validative Research (concept or idea testing in an attitudinal level) and Explorative Research (learning from users on a behavioural level). Through the Research Case Studies (Section 04: ‘Research in practice’), Validative and Explorative research methods are tested and accelerated to operate under the Agile work setting from Werkspot. The result is a Qualitative Research Process for Werkspot, through this process, the company can continuously involve users in the development of the platform.
... Read more

The construction industry is undergoing a significant change driven by the emergence of new technologies. Digitalization is transforming the sector with the Internet of Things (IoT) being one of its core enablers. As a leading multinational organization that provides cutting-edge tools, technologies, software and services for the global construction sector, Hilti is responding to this change by shifting from a product-oriented focus to become a more solution-oriented company. One essential factor of this transition is the adoption of the IoT technology to provide new solutions to address the customers’ needs in the digital future. Thus, Hilti needs to identify and seize opportunities in the field of IoT. The purpose of this graduation thesis is to enable Hilti to identify new ideas in the area of IoT and protect them with intellectual property rights to enable long-term differentiation. A good overview of the IoT patent landscape will provide clarity to spot new business opportunities systematically. The project aims at developing an innovation process that captures ideas from within the company. A qualitative user research was conducted uncovering several pain points in the current process. The loss of ideas and the lack of quality of ideas were found to be the biggest challenges that are hindering innovation from happening. The findings were contrasted with literature, which pointed out the importance of the development of an abstract idea to become more concrete, in order to derive value from or protect it.
In an iterative process, a solution was developed that addresses the identified pain points. The outcome is a combination of a sequential and iterative innovation process structured into four phases, including ideation, collection, assessment and protection of ideas. The basic concept of the process is the validation of assumptions, for which a tool was developed that facilitates this and acts as the main touchpoint throughout the process. The process is visualized in a process blueprint providing detailed information about the different components and interactions. An implementation plan was developed, giving suggestions for the next steps to be taken to operationalize the process.
The project provides a feasible solution for Hilti to identify and protect new ideas in the area of IoT. For the implementation, it is recommended to give much attention to the guidance and training throughout the process to ensure its correct use. ... Read more