Behavioral Signals provide AI-mediated conversation for contact centers. Automatically they “match each customer to the best-suited agent using voice data and emotion AI”. We collaborated to solve a company challenge that would help the developers that work on AI products use their APIs: Designing a web app that will allow developers to test out the API for emotionAI.

THE PROJECT

Team: 2 Designers
Duration: 4 weeks
Goal: Make the technology of Behavioral Signals accessible to more and more developers
Target Users: Users with sleep Issues, Users familiar with health/wellness apps that are curious about their sleep habits
Tools: Figma, Balsamiq
Deliverables: User Research, Information Architecture, Interaction Design, Visual Design, Prototype

the business needs

We didn’t lose much time and started planning our next steps. We organised a kick-off workshop and invited the VP of Engineering in Behavioral Signals, Nasos Katsamanis. We aimed to learn more about the company, the product, and values. It was crucial to understand the company’s needs and the value proposition of its product and further investigate how they could better address their users’ challenges.

Although the technology was accessible to developers through OliverAPI (a tool of Behavioral Signals), they were able to try it for free and the business needed to start monetizing the access requests. On the other side the API users had to deal with three major challenges: a. the delayed access after their request was submitted, b. the extended documentation that they were not reading c. and the lack of proper UI. They had to wait for hours to get access and then they were not able to easily understand how the technology works and interpret the outputs of the API.

We had all we needed to initiate the business research and to conduct a feature competitive analysis for all the direct competitors of Behavioral Signals: Deep affects, Webempath.net, symbl.ai, audeering, and vernai.com were the most important competitors with AI technology.

the users

Talking with developers that are experienced with API was very enlightening, as we were not that familiar with the challenges of this technology. Understanding the developers’ struggles when building products with APIs and discovering their preferences would help us to make their work easier. We conducted User Interviews with 5 developers. Reading and understanding the API technology was a common problem for all the developers and for most of them, it was important to be able to understand what the errors during the process would mean so that they could fix them.

problem and Hypothesis

It was clear that the main problem for the developers was that they could not use easily the OliverAPI tool so they can start building emotionAI products. Based on the research and the User Journey that we mapped out, we made the Hypothesis that by designing a WebApp with a simple UI and clear documentation the developers would be willing to pay for API that they could use in their emotionAI projects.

IDEATION

We started ideating on the MVP that we could design within the next couple of weeks.

The main elements of the MVP would be Engagement, Visualisation, and Documentation

Visual Design

Based on the users’ feedback and the product’s values we decided to focus on four Brand attributes : AI, Future, Idea, and Emotions. Using the OliverAPI logo colors to highlight the application’s functionalities, we created a white background dashboard, that would help the busy developers find what they want without distractions. The colors mainly would help them interpret easily the data graphics and would guid them into completing the task faster.

The solution

After two rounds of Usability testing with 5 developers, we finally delivered after 4 weeks a high fidelity prototype as the MVP of a new API web application .

See how Panos, the Software developer working on EmotionAI projects, finds the needed API documentation and learns how to the Emotion analysis works. Follow him when he checks his remaining balance and when he navigates in previous projects, and observe how he uploads new audion and checks the API usage data.

Conclusions

We were glad to see that the Behavioral Signals stakeholders were happy with the suggested solutions and were eager to start developing it.
From our side, we enjoyed working on this new challenge, especially because it was related to such an innovative a tool. We learned a lot and specifically it became very clear to us how important business and user research is to better understand the needs of the users and the business challenges. We also practiced a lot our tiime management skills and our flexibility when working on projects with more stakeholders, that have various visions and expectations.

If we had the opportunity to keep improving the application, in our next steps we would
– enrich the Audio Analysis page with more emotional data
– integrate the documentation page with other features
– create interactive dashboards
– improve the design by keep conducting more usability tests

Despite the challenges, finding solutions to uknown problems of new users was a great experience and I would like to thank my colleauge Rosa Chiacchio for the once again great collaboration!
Special thanks to Nasos Katsamanis and Behavioral Signals, for trusting us and for giving us the opportunity to work on this state of the art project.