Samsung Germany & Ekipa: Solve for Tomorrow 2024
This project was initially a pitch for a contest held by Samsung Germany and Ekipa. My team was not selected, but I felt incredibly proud of our efforts to not showcase it. Below is an elaboration of the submission and some visual examples.
This project demonstrates my ability to reflect on a social issue, imagine a technological solution, research to support the idea, conduct interviews, analyze and translate the findings into a design strategy, manage the project progress through collaboration and oversight, and take a setback and feedback with stride and go back to the drawing board. This last point will be briefly discussed at the end in "What's next?".
Team Saras:
Drishti Sanger (Researcher, Linguist, Project Manager & Product Designer).
Lucille Coats (Anthropologist, Graphic Designer & Video Editor).
in collaboration with: Masudur Rahman (UX/UI & Chatbot Designer).
Our greatest and most often overlooked asset is time. When placed against the backdrop of bureaucracy, especially government job agencies, it becomes something that is expected to be given. Our current research demonstrates there is growing frustration with command-based chatbots and a lack of effective interactions with agents due to time and energy constraints.
With the understanding the AI cannot solve all our problems, we considered the design of the context the product would be used in. Therefore, Project Saras is a means to bring out the potential of government job centers through:
1) Deploying a generative AI chatbot that converses with users.
2) Reworking the current methodology so that these spaces reflect a mentorship rather than solely bureaucracies.
The German job agency was considered as a starting point due to the client being Samsung Germany. Therefore, the German agency's command-based chatbot and website were used as starting points for our model (reference: https://www.arbeitsagentur.de/).
Brand development by: Drishti Sanger
Logo iterations by: Lucille Coats
Tools Used: FigJam & Zoom Meetings
User Profile was created from an interview with the participant. They were asked to speak about their experience with the job agency and then share their screen to browse their site/chatbot. The chatbot features I designed were based on these interviews and users pain points with the existing command based chatbot.
Interviewed by: Drishti Sanger
Tools Used: FigJam & Zoom Meetings
Interviewed by: Drishti Sanger
Tools Used: FigJam & Zoom Meetings
Interviewed by: Drishti Sanger
Tools Used: FigJam & Zoom Meetings
Interviewed by: Lucille Coats
Tools Used: FigJam
To start developing the features of the chatbot by visualizing the channels.
Sketched by: Drishti Sanger
For the visual elements, we dreamt of rich floral tones that evoked inspiration, and complimented the logo, but did not clash with the government website’s color scheme.
Brand development by: Drishti Sanger
Designed by: Lucille Coats
Website UI: Masudur Rahman
The lotus has the incredible ability to flourish in muddled waters. Similarly, the purpose of this project was to make sense of data collected through chatbots through qualitative research methods. Ultimately, delivering tangible results for users beyond the screen.
Brand development by: Drishti Sanger
Designed by: Lucille Coats
Tools used: Figma & Claude
The conversation component was created by interviewing participants to understand their needs and motivations, designing a sketch of a conversation flow, utilizing Anthropic’s Claude to test a potential incident. Lastly, Claude's output was refined to design the conversation for Saras.
Conversation script & design by: Drishti Sanger
UI prototype designed by: Masudur Rahman
Brand design by: Lucille Coats
Showcasing the inclusive language feature of a Generative AI chatbot as opposed to limited linguistic options on a command-based chatbot.
Tools Used: Figma
This feature was interesting to design as a chatbot feature because of its simplicity. From the interviews, I learned about the tedious task of browsing through the categories to set your parameters. So I asked, "What if the chatbot did that for you? I mean you are already immersed in the conversation, so why not?".
Another issue I wanted to address with this feature were issues it encountered while processing the request. For example, it notes that only a handful of jobs were found in that field. It would be helpful for the user to know beforehand. This is not a reflection on the chatbot, but on the database and this information could be useful for job agents as it can help them determine what positions are in demand.
Prototype & Video Credits:
Music & Edits by: Lucille Coats in collaboration with Masudur Rahman (prototype walkthrough video).
The UI prototype designed by: Masudur Rahman
Conversation script by: Drishti Sanger
I am integrating this project into my Master's thesis to better understand how ethnographic methods can to interpret chatbot data from an experience based approach. This is particularly fascinating when we think about artificial intelligence and how models are taught meaning.
The methods I am planning to utilize are: participatory mapping, narrative interviews, and analyzing the transcript through ethnographic coding.