A Research Reflection on Embodied Safety Interfaces for Human–AI Futures

kivi: A Research Reflection on Embodied Safety Interfaces for Human–AI Futures

 is an experimental object that explores how physical, low-arousal interfaces can function as 

 in high-stimulation, algorithmically mediated environments. Rather than optimising productivity, engagement or performance, Kivi is designed to stabilise the human nervous system through embodied, non-verbal interaction.

This project originates from a central question:

Contemporary digital systems are optimised for:

These dynamics create chronic sympathetic activation (stress-dominant states), which can lead to:

kivi addresses this by proposing a different design primitive: 

Instead of representing information, kivi modulates 

kivi deliberately avoids screens, language and symbolic instruction.

This is based on embodied cognition research suggesting that:

 can become a primary interface for safety.

kivi: a quiet companion for nervous system

Developing the skills of a product designer, learned how to leverage AI and work with expert mentors to prepare for a lasting career in tech. 

Researched & collaborated on UI tasks on VR Gaming

& blockchain with artists, designers and developers.

Tools: HTML5, CSS3, Figma, Figjam & Adobe CC

Outcome: redesigned onboarding screens UI, improving user comprehension in usability tests from 60% to 85%.

Applied design thinking principles to human-centred AI projects and designed user interactions around AI features.

Tools: Scikit Learn, PyTorch, TensorFlow, Colab & Keras

Outcome: Entered Kaggle competition & came in Top 100.

, the Stage 2 Leader of BA Graphic Communication Design and MA Graphic Communication Design. 

Explored the topic "Activism of Design" together with Stage 1 students, My duty involved:

Designed icons, came up with concepts and illustrative ideas for denkwerk’s new partnership.

- Support in design work and visual brainstorming for various projects and pitches

- Independent design supervision of projects and clients

- Maintaining, performing website updates

- Facilitating co-operation between the design and technology teams

- Project work focused on profitability and compliance with objectives

In this role, the following clients were served:

Telekom, True Fruits, Lorenz(NicNacs, ErdnussLocken, Crunchips and Saltletts), OBI, NetCologne, MotelOne

I was given tasks to design both creative, conceptual illustrations and animations for several denkwerk internal projects as well as occasionally leading visual concepts and identities for some projects, which are as follows:

- Taking full responsibility and leading the creative aspects of Cologne Service Jam; a non-profit event, sponsored by denkwerk, the event involved 48 hours to develop new services from idea to prototype, under one global theme, offering the opportunity to get hands-on experience on service design methods, gathering creative thinkers and having fun.

- Featured Illustration for The t3n - digital pioneers magazine, denkwerk #ReLab pitch article on the topic “self-disruption for the agency of the future”.

Not a debate. Not a class. Just a place where thinking is allowed to happen.

Concept Design · Human-centred Learning Environment · AI-assisted Systems

 is a conceptual design project exploring how digital environments can support thinking, learning, and dialogue without hierarchy, performance pressure, or competition.

The project was born from a simple question:

What would a learning space look like if it prioritised psychological safety, curiosity, and presence over speed, productivity, or persuasion?

Thought Playground is not a social network, a debate platform, or a gamified learning tool.

 designed for slow, respectful inquiry; a place where people can explore ideas together without needing to perform, convince, or compete.

The space is intentionally non-hierarchical:

The aim is to create conditions where thinking can unfold gently, rather than be forced.

The visual and interaction design draws inspiration from:

Participants are represented by softly rounded, white, ambiguous forms; intentionally without facial features, gender or customisation.

This removes visual hierarchy and reduces comparison, allowing attention to remain on shared inquiry rather than self-presentation.

Movement is slightly imperfect and playful, encouraging ease rather than mastery.

The environment uses soft pastel tones and minimal contrast to support nervous-system regulation and sustained attention.

Sound, when present, functions as an ambient environmental layer rather than content; supporting calm rather than demanding attention.

Thought Playground is designed with explicit ethical boundaries:

The project explores how AI and interface design can 

 rather than dominate, allowing human thought and connection to remain central.

Thought Playground exists as a speculative yet practical exploration of:

human-centred systems, ethical AI, and environments that shape how people think rather than what they think

The decision to represent users through childlike, non-specific forms in this experiment is grounded in findings from cognitive neuroscience, developmental psychology and affective science. Research shows that simplified, ambiguous representations reduce social threat perception by bypassing adult status cues such as dominance, competence, gender or attractiveness, which are known to activate comparison, self-monitoring, and performance anxiety in the prefrontal cortex. 

In contrast, childlike forms tend to engage neural pathways associated with caregiving, play and exploratory learning, particularly within the limbic system, fostering a sense of safety rather than evaluation. From a developmental psychology perspective, early childhood is characterised by learning through play, curiosity, and relational attunement rather than outcome-driven performance; abstracting users into childlike silhouettes reactivates these modes of engagement without infantilising them. 

: gently lowering cognitive defensiveness, reducing identity fixation and allowing participants to inhabit a more open, exploratory mental posture. By removing markers of adult identity while preserving agency, the environment encourages presence, mutual regard and shared inquiry; conditions shown to support deeper reflection and collaborative thinking across both neurotypical and neurodivergent populations.

Thought Playground — Design Principles (Derived from Risk)

Thought Playground: a space where ideas can rest, collide and change

Designing a lightweight gamified AI Banking App for young people struggling with money management.

PlutoPay: Modelling Financial Behaviour as a Cognitive-Emotional System

For many young adults in the UK, interacting with personal finance tools has become emotionally aversive rather than empowering.

43% of 18–27-year-olds avoid checking their bank balance due to money anxiety

, creating a negative feedback loop: anxiety drives avoidance, while avoidance worsens financial and mental wellbeing.

PlutoPay treats money management as both a 

Instead of framing anxiety as a behavioural flaw, the design acknowledges it as a 

state that can be buffered and reshaped through interaction

UX flow design, gamification system design, prototyping, usability testing facilitation

Financial behaviour cannot be meaningfully changed without addressing its 

Making behavioural structures visible supports meta-cognitive engagement, enabling users to interact with finances more intentionally.

Exploring adaptive systems that respond to emotional state shifts over time.

PlutoPay: AI Money Managing Banking App

Designed for therapy waitlists: a private, calm AI for emotional reflection and therapeutic skills, prioritising trust and user autonomy.

Pastel: An Interface for Emotional Reflection

Access to mental health support is increasingly limited, with long therapy waitlists leaving many individuals without timely emotional support.

During this gap, people often struggle to articulate emotions, fear being judged, and lack psychologically safe spaces for reflection. While AI-based companions are emerging, many prioritise response generation over reflective support, risking emotional dependency or superficial reassurance.

How might we design an AI-mediated interface that supports emotional reflection while preserving user autonomy, ethical boundaries, and psychological safety?

Pastel was designed not as a digital therapist, but as an 

interface for externalising and navigating internal emotional states

 rather than an authority—supporting users in organising thoughts without directing conclusions.

Ideation explored how subjective emotional experiences could be 

represented without being quantified or judged

UX flow, interaction design, ethical AI framing, high-fidelity prototyping, usability testing

Emotional support systems are less effective when they attempt to simulate empathy, and more effective when they support 

Designing for wellbeing requires treating emotions not as data points to resolve, but as 

experiential structures users need help navigating at their own pace

Pastel: Your Ethical AI Emotional Companion

Ky’s design decisions are insightful, intuitive and well-reasoned.

Noah Hassan

Ky’s attention to detail and ability to balance aesthetics with usability stood out in projects we collaborated on.