The first phase of the "HAUS UPDATA" project, conceived and planned by Dentsu Inc. to verify "valuable data utilization that aligns with residents' feelings," was conducted from May to October 2024. This project promotes smart home adoption by collaborating with housing manufacturers, appliance makers, and sensing service providers. It involves installing various sensors and IoT devices in participants' homes to facilitate the utilization of lifestyle behavior data and support improved living.
While the first part covered the results of the first phase, this second part features an interview with Erina Tanaka and Hayato Tamura from Dentsu Data & Technology Center, and Mayuko Takai from Dentsu Digital Inc. about the outline and key evolutions of the second phase, scheduled to begin in summer 2025.
"A Home Supporting Life Balance" – Returning Data to Residents
Tanaka: In the first part, we shared the findings from the first HAUS UPDATA pilot project and feedback from our partner companies. By linking the collected lifestyle data and providing recommendation information to support better living, we achieved high satisfaction levels among both the monitor households and our partner companies.
Therefore, starting in summer 2025, we will conduct a second-phase demonstration project themed "A Home Supporting Life Balance" over a six-month period. We are expanding the number of monitor households from 10 in the first phase to approximately 100 and have secured cooperation from more partner companies. Furthermore, we plan to provide monitor households with dedicated tablets. By delivering recommendations based on their lifestyle data—such as tailored product usage guidance and living advice—we aim to provide valuable services while maintaining communication with residents. Mr. Tamura, who leads data analysis, how do you plan to analyze lifestyle data and deliver valuable information to residents?
Mr. Hayato Tamura, Dentsu Inc.
Tamura: I want to visualize the analysis results of lifestyle activity data using dashboards and reports to clearly communicate them to our monitors and partner companies. Lifestyle data collected via motion sensors and similar devices is often complex, noisy, and challenging to analyze and process. Furthermore, with both the number of monitor households and partner companies increasing this time, the volume of data logs will be enormous, making processing even more demanding. However, I believe tailoring the data into easily understandable formats tailored to each partner company's specific needs will generate greater value.
Furthermore, while the first phase had 10 monitor households, this phase will expand to 100 households. We anticipate this will enable comparisons between households. As the number of monitor households grows further, we believe we can also obtain statistical data showing broader household trends within Japan.
Developing an AI chatbot that supports daily life through both assistant and coaching functions
Tanaka: For the second phase, we are currently discussing how to interpret consumers' lifestyles and latent needs from their behavioral data, and how to provide information and services that make consumers feel the benefits of sharing their data. Because the data is so closely tied to daily life, one potential application we see is capturing changes in daily rhythms that people aren't consciously aware of, and using that to help them regulate their routines. Dentsu Digital Inc. is currently developing an AI chatbot for this purpose. What are its key features?
Takai: This AI chatbot has two main roles: Assistant and Coaching. The Assistant is a chatbot that engages in casual conversation with users and answers their questions.
The coaching function, on the other hand, is a chatbot that provides specific advice based on each individual's data regarding small goals they have set for themselves. For example, if a user sets a goal like "I'll go to bed early and wake up early this week," it might say things like, "You got home early today, so you can go to bed early," or "Great job waking up early!" It evaluates actions related to the goal. Furthermore, we are designing it to be capable of coaching that offers new suggestions, such as, "Last week your goal was this, so how about trying this new challenge this week?"
Ms. Mayuko Takai, Dentsu Digital Inc.
Tanaka: When we first discussed the AI chatbot with you, Ms. Takai, I was struck by your comment about creating an AI chatbot that feels like a member of the family. What role does this AI chatbot aim to play within the household?
Takai: Even within families, some members might eat separately, and some households have limited communication. We hope that by having the chatbot initiate greetings and conversations, it can spark more dialogue within the home. Currently, we plan to provide simple instructions like "use this style of conversation for households with children," but eventually, we envision it being able to tailor the information presented and conversational style to each individual family member.
We also want to collaborate with caregiving and pet services to support better living.
Tanaka: In the first phase of our pilot project, we partnered with various companies to build a foundation for delivering new value based on lifestyle data. For the second phase, we aim to make the services built on this foundation even more valuable.
Also, while support for life balance and daily rhythms often focuses on diet and sleep, we want to tackle broader challenges going forward. For example, if we can collaborate with services related to caregiving and pets, I believe we can improve the lives of even more people. Tamura-san, from your perspective on data utilization, and Takai-san, from your perspective on AI chatbot development, do you have any outlooks?
Tamura: Through this project, I'd be delighted if we could grasp lifestyle trends in Tokyo and other regions. I hope we can increase the number of monitors while improving the accuracy of our lifestyle log data analysis.
Dentsu Inc. - Erina Tanaka
Takai: At Dentsu Digital Inc., we've handled AI chatbot projects before. By taking on challenging projects like this one, where we develop AI chatbots with diverse roles using various multi-agent systems (*1), we hope to differentiate ourselves from other companies' AI services.
The second phase of our proof-of-concept experiment aims to significantly increase monitor households and update lifestyles through the return of consumer data. Expectations are high for this advanced project, featuring detailed analysis combining multiple lifestyle behavior data points and data-driven coaching.
※1 A system where multiple AI agents interact and collaborate to solve specific tasks or problems.
Previously worked in the data science department of a think tank company, responsible for analysis and AI model development using financial and economic data, primarily stock price data. Joined Dentsu Inc. in 2023. Engaged in data science tasks within the data cleanroom business, including analysis, machine learning model development, and generative AI utilization.
Since joining Dentsu Inc. in 2021, he has been responsible for verifying advertising effectiveness and developing solutions utilizing platform data. Concurrently, he has been engaged in data technology-related work, including supporting partner companies in data utilization, primarily within the IoT industry.
After five years in R&D at a major electronics manufacturer, joined an AI services startup in April 2021 as its first technical salesperson (also serving as PM). From 2021, led the new data utilization consulting business, working on-site at client locations to handle data analysis and utilization tasks. Joined Dentsu Digital Inc. in 2024, where I am engaged in proposing and driving Proof of Concept (PoC) projects for systems utilizing generative AI.
