Aiaibot
Aiaibot makes it easier for enterprise companies, automating their customer dialogues at all digital touchpoints and ensuring both - more efficiency in their processes and a higher quality of services.
Overview
Normally, the use of such Machine Learning algorithms and the training of classifiers require very specific know-how and expert knowledge. One of the goals in developing aiaibot is therefore to simplify this process and make it accessible to more people. Based on the feedback from our customers, I believe we have achieved this goal.
"The biggest challenge in the implementation of customer projects is the ability to provide high-quality training data in sufficient quantity."
This has a crucial influence on the results that our customers can achieve. To pinpoint the problem, we had to make sure that the user which had a model evaluated, can see and understand its performance. We wanted to visually guide the users through the process, and introduce them to more advanced key metrics, to get the full advantage of the tool.
The process
Different methods and rhythms have been tested to enable Agile Design within the company. Working in "Pre-Sprints", to do research and build the Information Architecture related to the new features, followed by an Agile Designing Phase which is iterated by Designers and Frontend Engineers jointly. While solving problems of recent Sprints and delivering high-quality designs to our developers, we always improved our setup and Design System, making sure our components were durable over time.
While we created a design style/language for the company, I also collaborated closely with the machine learning engineering team to break down complex ideas and communicate them visually in a clear and consistent manner. One difficulty was that many designs depend on the structure of the user data.
Moreover, our designs had to be thought through completely from the start because the front end was implemented by a third party and later changes were costly. That meant that I also conducted quality assurance and guided our overseas contractors. I believe that our part of the product has an appealing UI/UX that helps the users to achieve their goals.
UX Research, User Interviews & Personas
UX Research, User Interviews & Personas
Based on the experiences of our previous research team and the development of the on-premise solution lena by PIDAS, we had the chance to include their knowledge into a new mission and vision for a scalable and dynamic SaaS product: aiaibot.
Key findings:
We then focused on who is going to use our AI functionalities within the platform. In order to validate our software experience, we decided to create three different target groups: Agency, Self-Doer, and Enterprise.
Conducting user interviews and defining personas:
I conducted a focus group to learn about our potential users' approach to conversational AI and Natural Language Processing tools. Talking to real people helped me understand what causes the most problems while they are using AI functionalities on their working machine and what could be improved.
Reusable components
In order to ensure consistent user experiences across the different tools available on the platform, we kept our components up-to-date and documented in Storybook. A solution that helped designers & front ends to be on the same page and re-use components to increase efficiency.
The outcome
The outcome is a powerful, easy to use and simple to integrate platform that allows users to efficiently train a state-of-the-art multilingual machine learning model on their own data with just three clicks.
I conducted quality assurance on overseas third parties contractors working on Github, making sure that the delivery was as expected.
It was a great challenge and I'm really grateful for all the feedback we got from users over time. Thanks to that, we made sure to shape the product on their needs, increasing the success rate on their tasks.
