Ally Assist -
Conversational AI
In today's fast-paced world, sleep has become an increasingly rare and precious commodity. Poor sleep can have a significant impact on both physical and mental health, leading to decreased productivity, mood swings, and a host of other issues. To address this problem, Philips tasked my team with creating a mobile app that will help people actively engage with and improve their sleep. Our app will provide users with the tools and resources via a conversational AI to help track their sleep patterns, set goals, and make lifestyle changes that can promote better sleep hygiene. By leveraging the latest technology and cutting-edge research in sleep science, our app empowers users to take control of their sleep.
Background
-
Philips sleep business SRC lacks any real presence in sleep conditions that is not apnea.
-
Sleep conditions are on the rise and Philips needs to have a scaleable solution to meed the demand
The challenge
-
Everyone sleeps but not everyone gets the right kind of quality sleep
-
Sleep conditions are complex and require a multi disciplinary support
-
Sleep specialists are rare, and oftentimes overwhelmed at clinics
Project details
-
Role: Creative Lead
-
Team: 12
-
Budget: 800K
-
Timeframe:
-
Testing: 2 Months
-
Build: 8 Months
-
00. Ethical principles for healthcare AI
Ethics in healthcare AI ensures that the development and use of AI technologies adhere to ethical principles such as patient autonomy, beneficence, non-maleficence, and justice. This is important to prevent harm to patients, maintain their trust in healthcare providers, and ensure that the use of AI technology in healthcare is transparent, fair, and respectful of patient rights and privacy.
01
Protect autonomy: Humans should have oversight of and the final say on all health decisions — they shouldn’t be made entirely by machines, and doctors should be able to override them at any time. AI shouldn’t be used to guide someone’s medical care without their consent, and their data should be protected.
03
Ensure transparency: Developers should publish information about the design of AI tools. One regular criticism of the systems is that they’re “black boxes,” and it’s too hard for researchers and doctors to know how they make decisions. Philips wants to see enough transparency that they can be fully audited and understood by users and regulators.
05
Ensure equity: That means making sure tools are available in multiple languages, that they’re trained on diverse sets of data. In the past few years, close scrutiny of common health algorithms has found that some have racial bias built in.
02
Promote human safety: Developers should continuously monitor any AI tools to make sure they’re working as they’re supposed to and not causing harm.
04
Foster accountability: When something goes wrong with an AI technology — like if a decision made by a tool leads to patient harm — there should be mechanisms determining who is responsible (like manufacturers and clinical users).
06
Promote AI that is sustainable: Developers should be able to regularly update their tools, and institutions should have ways to adjust if a tool seems ineffective. Institutions or companies should also only introduce tools that can be repaired, even in under-resourced health systems.
01. Discover
It was necessary to explore the scope and value of sleep apps with users. This involved gathering feedback from individuals who have used sleep apps before or who were interested in using them to better understand their needs and preferences. Once the scope and value were established, it was important to select the type of AI conversation that will best meet their users. This could include chatbot-style conversations, voice-activated interactions, or a combination of both. By taking these steps, we could create a sleep helper AI that is both useful and user-friendly, ultimately helping individuals achieve better sleep and improved overall health.
Understand what users want from a sleep helper AI
Explore the scope & value of the app with users
Selecting the type of conversation needed
02. Define
It was crucial to consider the needs and preferences of the users who will interact with the sleep AI. One way we did this is by creating user personas, which were fictional representations of the app's target users. These personas provided a clear understanding of the users' goals, challenges, motivations, and behaviors, allowing us to design an app that caters to their specific needs. In addition to user personas, creating a sleep AI app persona is also essential in the define stage of a project. This persona represents the app itself, giving it a unique personality, voice, and tone. It allows developers to create a consistent user experience and develop the app's communication style. Together, user personas and app personas provide a foundation for designing a sleep AI app that meets the needs of its users and is easily identifiable as a brand. By taking the time to create these personas in the define stage, we ensured that our app is effective, user-friendly, and memorable.
Personas | Users
Personas | The consersational AI persona
03. Develop
It was important that we created a well-designed conversational flow that provided a seamless user experience. Conversational flow diagrams are a powerful tool that can help this.These diagrams detailed the paths users can take during their conversations with the app and the responses that the app will provide based on user inputs.Conversational maps were another important step in the process of creating a conversational AI application. Once the conversational flow has been designed, developers can create a detailed map of the conversations, specifying the exact words and phrases that the app will use during each interaction. This allows developers to ensure that the app's responses are consistent and on-brand, making for a better user experience. Finally, behavioural change programs were integrated into the conversational AI application to promote positive changes in users' behavior. These programs use the app's conversational flow to encourage users to adopt new habits or make healthy lifestyle changes, making the app more than just a tool for information retrieval. By using conversational flow diagrams, conversation maps, and behavioural change programs, developers can create conversational AI applications that are engaging, effective, and help users achieve their goals.
Conversational planning
Conversational flow diagrams for all journeys
Conversations mapped for development
Behavioural change program draft
Early wireframes
When creating wireframes for a sleep-assisting conversational AI, it's was important to consider the user's journey through the app. This includes the initial greeting and onboarding process, the selection of a sleep aid feature, and the provision of personalized sleep advice or suggestions.It's also important to consider the app's visual elements, including typography, colors, and imagery, which can have a significant impact on the user's engagement and trust in the app. We ensured that the app is designed with user needs in mind, leading to a better overall experience for the app's users.
Using data to drive app decisions
By collecting and analysing data on user preferences and behaviours, we could determine whether users prefer a more personalised, conversational approach or if they prefer the convenience of a connected sleep app. They can also assess whether users respond well to gamification elements such as rewards, challenges, and progress tracking.
Ultimately, the choice between these approaches depended on the target audience and the app's overall goals. For instance, if the app is designed for individuals who struggle with insomnia, a sleep coaching conversation may be the most effective option. If the app is designed for a broader audience, a connected sleep app with personalised insights and recommendations may be more appropriate. Gamification elements can be used to incentivise users to make positive changes in their sleep habits.
In conclusion, data-driven decision-making helped our team & stakeholders determine the best approach for a sleep-focused application. By gathering and analysing data on user behaviour and preferences, we made informed choices to create an effective and engaging sleep-focused application.
Coaching conversation
Connecting other apps
Gamification of sleep
Creating visual treatments | User accectence testing (UAT)
Different visual treatments can significantly impact how users perceive and engage with the app. Therefore, it was important to consider various visual options and test them with users to determine the most effective approach. Designing different visual treatments for our sleep app involved experimenting with colour schemes, typography, iconography, and other visual elements. For example, some designs emphasised soothing colours and calming imagery to promote relaxation, while others used bright colours and bold typography to communicate energy and vitality.
To determine which visual treatment was most effective, it was important to conduct user testing. This involved presenting users with different versions of the app and collecting feedback on their preferences, ease of use, and overall satisfaction. By analysing user feedback, we could identify the visual treatment that resonates most with users and creates the most positive experience.
Conversational testing
The goal of testing our conversations with a sleep AI was to ensure that the app's interactions are natural, helpful, and engaging. This involved testing different conversation flows and ensuring that the AI responds appropriately to user requests and questions. We also wanted to ensure that there were no conversational breaks in doing a wizard of OZ testing method with the pre-made dialogue flows. Through testing conversations with our sleep AI, we identified areas where the AI may be confusing or ineffective and make improvements accordingly. This feedback informed changes to the app's dialogue, its interface, or its functionality, leading to a more effective and engaging user experience.
Conversational planning
Conversation test 1st draft
Conversation test 2nd draft
Conversation test 5th draft
04. Deliver
05. Development testing
During the testing process, we used a variety of tools & techniques, such as automated testing software & manual testing. Automated testing helped identify bugs and errors in the code, while manual testing helped identify usability issues and other user experience problems. To ensure that the testing process was effective, we established clear testing goals and objectives, created test cases that covered all aspects of the app's functionality, and document any issues or bugs that were discovered. This documentation was used to inform future development efforts and improve the app's overall quality.
Real User Trials
Real user trials involved us soliciting feedback from a group of real people with real sleep issues who represented the app's target audience. To conduct these real user trials, we recruited a group of users and provided them with access to the app. During the testing process, we asked them to perform a variety of tasks using the app, such as setting a sleep goal, tracking their sleep habits, and engaging in conversations with the AI. Feedback should be collected during their interviews, and other feedback mechanisms to identify any usability issues, bugs, or other problems that may have been missed during development.
The verified solution
The final solution for Ally Assist, a conversational sleep AI, represents the culmination of a comprehensive and iterative design process aimed at delivering an effective and user-friendly sleep assistant. The solution incorporates a range of features and capabilities designed to help users improve their sleep habits, including personalised coaching, sleep tracking, and AI-powered conversations. The user interface for Ally Assist was designed to be intuitive and easy to use, with clear and concise messaging and visually appealing graphics. The app guides users through the process of setting sleep goals, tracking their sleep habits, and engaging in conversations with the AI to receive personalised coaching and advice. The AI conversations are designed to be natural and engaging, with contextual responses that address users' specific needs and concerns. The app also features a wealth of helpful articles and insights, providing users with additional resources to help them improve their sleep habits. To ensure the effectiveness of Ally Assist, real user trials were conducted, with feedback gathered from a group of real users who represented the app's target audience.
06. Impact
Ally Assist, a conversational sleep AI, has the potential to make a significant impact on users' sleep habits and overall health and wellness. By providing personalized coaching, sleep tracking, and AI-powered conversations, the app helps users improve their sleep habits and achieve better rest and recovery, as detailed in the following impact information.
Feeling empowered to handle their sleep rose from 6.15-7.5
-
These figures are based on a scale from 1-10 (Details above)
-
The mode of the app rating was at an 8 out of 10
Nearly all users felt understood and in control
-
Overwhelmingly conversations resonated with users with almost all participants having positive and helpful conversations with information they could use.
75% of users got accurate
recommendations
-
Most users found the reccomendations useful, for most the reccomenndations were not new information.