1st Place Winner at the Springboard Hackathon 2023. Click here to see why we won.
Role
Deliverables
Tools
Team
Pathos AI is an artificial intelligence (AI) application designed to seamlessly integrate with messaging platforms such as Slack. It enhances empathy and emotional intelligence in remote work environments by suggesting appropriate messages tailored to the tone of the conversation, based on chat history.
Here is a demo showcasing the final product:
To gather insights, we created and distributed a survey to remote workers, focusing on their usage of AI and chat applications, as well as their communication experiences with colleagues. Here are the key findings:
ChatGPT
The majority of respondents favor ChatGPT as their AI tool of choice
Slack
A significant portion of participants regularly uses Slack for their professional communication
Communication
Although a subset reported negative experiences with communication and company culture, the majority expressed positive sentiments
Informed by the valuable insights gathered from the survey, it seemed a wise decision to design a plugin to:
1
Enhance the productivity by eliminating the need to switch between Slack and ChatGPT.
2
Aim to bridge communication gaps by fostering emotional intelligence and empathy.
We initiated a comprehensive study of comparable competitors, gathering inspiration and strategic insights. We explored companies such as Grammarly, Mei, Personal AI, ProWritingAid, LanguageTool, Cresta AI, and ContactOut. Here are the 'Top Takeaways' from our analysis:
Based on the research, we began creating user flows for two scenarios:
1- When there is chat history, the AI suggests the best messages based on the top three tones derived from that history.
2- When there is no chat history, users can select the tone first and then write their messages to receive suggestions from the AI.
Here is the user flow when there is chat history:
Upon finalizing the user flows, we translated conceptual ideas into visual representations through the creation of wireframes. These wireframes served as a clear roadmap, guaranteeing a cohesive and engaging user experience.
After creating the wireframes, we conducted a usability test to determine how easily users could utilize the AI suggestions and gather insights into their opinions.
Participants:
- 5 remote workers
Here, you can observe "Before" and "After" screens. The "Before" screens correspond to the wireframes, while the "After" screens represent adjustments made after conducting the usability test. As we collected users' feedback during the test and refined the screens following the creation of high-fidelity versions, the "After" screens are presented in high fidelity.
Finding 1: Struggling to connect with icons, disliking the 'Report' option, and desiring regeneration of suggestions.
Finding 2: Having difficulty locating Standard and Auto Draft options.
Finding 3: Desiring to be writing a prompt.
Finding 4: Aspiring to select a tone.
One challenge we faced was determining whether we should receive suggestions from the AI for the entire paragraph or for each sentence separately. We opted to conduct A/B testing following the usability test to gauge user perspectives and identify the most effective solutions aligned with their goals.
As per the results of A/B testing, users showed a preference for receiving suggestions for the entire paragraph.
After creating the High-Fidelity Prototype, we conducted a supplementary usability test to assess users' ease of interaction with refined AI suggestions based on insights gathered in Usability Test 1 and A/B testing.
Participants:
- 5 new remote workers
While users expressed satisfaction with the AI's overall user interface and provided positive feedback during the test, they also observed that the abundance of screens can be somewhat confusing.
Here is the Figma Prototype:
The Crucial Role of Usability Testing:
Helen Keller: “Alone we can do so little; together we can do so much.”
