A spotlight on AI in the care economy
It’s everywhere during these early summer days of 2023: news that the artificial intelligence (AI) revolution has arrived and, at The Holding Co, we’re actively researching the range of ways that AI is already and may impact care in the future.

At the intersection of AI, care, and design, we see a world of new possibilities for innovators and investors. So much of care work is administrative drudgery–the seemingly never ending and often largely invisible cycle of coordinating, navigating, scheduling, organizing etc. This cognitive and logistical load buries caregivers–unpaid and paid–on a daily basis. AI presents a profound possibility to lift the administrative burden off of caregivers’ shoulders.

Additionally, AI opens up new possibilities for tending to the safety concerns that are an inevitable part of caregiving. Many care companies are looking to AI to detect disruptions in typical user patterns, track health data via wearables, and offer new forms of interaction to cut down on isolation–all of which enables elders to maintain financial security and independence longer. Right as AI becomes more ubiquitous, 10,000 Baby Boomers a day are turning 65; there’s no question that this new technology will transform aging in the coming years.
Like so many others, we are also worried about this breakthrough moment in technology. There are some real fears of how AI might impact the care system, which is already broken in so many ways.

How do we leverage AI to handle the administration and safety concerns of care so we can devote our precious energy and time to the more human aspects–memory and meaning-making, connection, love?

That’s a question we will be answering–alongside and in thought partnership with our diverse network–over and over again together in the coming months and years. This chapter is just the first step.
five areas where ai is already transforming care
Many of those building technology in the care space are already experimenting with using AI to lighten the burdens of care and make operations more efficient. Check out these five functions we are already seeing AI play, along with corresponding case studies, and get inspired:
Care Concierge
In the last few years, a new role in the care system was born: care concierge. These experts–think social worker meets project manager–help families experiencing either acute or longer term health crises figure out how to make it through.

AI promises to completely upend the role of care concierges. It is already freeing them up from some of the more mundane aspects of the job–reading through medical records in order to summarize the big takeaways, finding the best fit products and services for a given caregiving situation, generating a letter for an insurance company advocating to have a particular treatment funded etc. With AI’s assistance, care concierges have more time to do the more relational pieces of the job like coaching individual caregivers through having difficult conversations about money and death with their families, as an example.
Care concierge is being presented with a fascinating scenario in which we can learn how best to integrate AI and human interactions–giving our concierges more space for high touch experiences (sibling dynamics, counseling on decisions etc.) while increasing efficiency to reach even more people.

The approach and skill sets that care navigation requires may be shifting in the coming months and years as this integration of tech and human touch continues to take shape.

The capacity to have nuanced, emotional conversations, more than do research in complex systems, will be pivotal for care concierges.
A screenshot of Duos's homepage
DUOS is building a platform to connect older adults to the services they need to thrive. Their conversational interface uses a large language model to recommend the best services to older adults and their caregivers in a way that’s customized, intuitive, and as a result, builds trust. This large language model draws on an aggregation of every Medicare Advantage plan benefit, government program, and community resource. It then matches the right resource using machine learning models that draw from DUOS historical experience. So let’s say you’re an older person whose mobility is compromised – you might do a Google search for scooters, and come up with a range of search results, much of which isn’t available in your area or wouldn’t be covered by your health plan etc. If you make this request via DUOS, you’ll get search results that jive with your preferences, locations, and benefit coverage. Your needs are met faster and with more precision, which means you aren’t isolated or left waiting on a solution that seems out of reach.
A screenshot of Wellthy's homepage
Wellthy just launched their first in a coming line of AI features, “sentiment analysis", which has the power to help their care concierge team more deeply understand the feeling behind messages from their members. What that means is when a member using Wellthy messages their team, AI will immediately register the sentiment of that message and determine if it's positive, neutral, or negative. If someone is expressing frustration about a doctor’s appointment, an unexpected diagnosis, or anxiety over a healthcare bill – to name a few examples – their sentiment analysis AI tool flags this for expedient action, giving more context to their care team and boosting the efficiency of serving their members. As  Wellthy’s CEO Lindsay Jurist-Rosner recently stated, “AI will never replace the human connection and expertise that is part of Wellthy’s magic. But I’m so excited for the ways it will provide fuel for our teams, and excited for the doors opening at Wellthy today.”
Mental Load
Keeping track of doctor’s appointments, birthday parties, soccer schedules and grocery lists–what researchers call “cognitive load”--has traditionally been the responsibility of women, even those who work full-time. There has been a huge untapped market of exhausted parents, especially mothers, who are desperate for support that might streamline their lives so they can focus on being present with their kids, connecting with their aging parents, or tending to their own neglected restoration and delight.
AI just might finally provide the technological superboost needed to make these services scale. The multi-modal capacity of GPT-4 means that a busy parent could quickly upload a PTA announcement and get the important information marked in their calendar, send audio recordings of things to remember and seamlessly be reminded at all the right times, even get suggestions on what's for dinner with all the groceries loaded into their virtual cart. The mental load is huge and AI is finally parsing and organizing that mess like the extra brain so many have been dreaming about.
A screenshot of Milo's homepage
Milo gives busy parents an AI collaborator to be on point for the running of the daily family operations, in recognition of the fact that it is work needing to be “staffed.” Today, where 60% of couples are dual-career, Milo pairs AI with human experts (parents) to take on and do this job. Parents simply forward everything to Milo - school newsletters, soccer schedules, summer camp reminders and Milo deals with it - getting it to calendars or back as reminders, all via simple SMS that makes it easy for everyone to use.
A screenshot of Instacart
Instacart’s new plugin leverages AI to simplify the ever-present “dinner dilemma.” You tell ChatGPT–Instacart’s GPT-4 platform and one of the more widely available GPT-4’s today–what’s in your kitchen, let’s say, cauliflower, penne pasta, and romaine, and ask “What’s a kid-friendly meal I can make, and what else do I need?” ChatGPT can now create Instacart orders based on suggested meal responses, adding all of the necessary ingredients to the cart in just a few clicks. In other words, users can leverage ChatGPT in order to take advantage of Instacart’s catalog, which already spans more than 1.5 million products from over 1,100 retail outlets, with way less stress.
According to our own research, there is an opportunity to activate up to 54 million consumers who are interested in, but not yet paying for routine childcare, adultcare, or both, not to mention the range of products and other kinds of services on the market. Based on a May 2022 consumer survey of 3,110 respondents, representative from a U.S. census-representative sample across gender, age, race/ethnicity, and income, we discovered that the majority of the market–54% of caregivers–are open to paying for care services, but not yet convinced they should. We call them “the care cusp” and AI is about to jumpstart uptake like never before.
Caregivers will be able to get personalized search results unlike anything one might find with a traditional google search, including both demographic and disease profile information, but also nuanced preferences, concerns, and desires. Imagine a burned out adult caregiver for a mother with Alzheimers, in desperate need of respite care, accessing–within seconds–all the public and private options in her local area and specialized information about how she might pay for such services, plus recommendations on ways to talk to her mother about the forthcoming experience. Or think about a parent trying to figure out the most cutting-edge approach to orthodontia and affordable options in his area for doing right by his kid; within mere seconds he has become a more astute consumer and pushed the whole marketplace to get smarter and more sophisticated about how it is reaching and serving parents.
A screenshot of The Mom Project homepage
the mom project
The Mom Project aims to create more economic opportunity for moms through a variety of strategies, including providing companies with a pool of talented working parents to hire and insights on how to support them better once they do. To date, they have over 1.2 million talented professionals in their network and over 3,000 companies signed up to hire moms to drive their businesses forward. The Mom Project is building a “candidate fit” process powered by AI through a very intentional design process they call, “Be the Algorithm;” essentially, they studied their recruiters by methodically going step-by-step through the entire experience, and only then were they able to integrate AI in a way that felt ethical and useful to their overall mission. AI now takes a bunch of factors (relevant job experience, how long in previous positions, skills, remote, full-time, part-time additional element of fit, etc.) and produces a curated list of candidates, along with rationales for these matches. They are continuously thinking about how to give the constantly learning model cues about best fit, such that it will actually diminish bias as opposed to perpetuating it (some people’s fear around incorporating AI into human resources work).
A screenshot of shop.app/AI
Shop.app/ai allows users to enter a prompt—I’d like to buy my 75-year-old mom a pill organizer under $25, but she hates plastic. What are my options?—and instantaneously get image-rich options with all the essential information they need to make a quick, informed decision. One can imagine the future implications for this kind of product and service search for care consumers as marketplaces continue to evolve and AI becomes smarter and smarter about what works best for caregivers.
Life alert pendants, Nest cameras, and other remote technology have been used for years to help adult children take care of aging parents from afar. The integration of AI, however, is rapidly raising the bar on what caregivers can expect from tools that keep elders in their homes longer and under far safer conditions. Wearables or devices installed in the home that detect motion can now collect massive amounts of data, that data can be instantaneously analyzed to infer actions or patterns in activities of daily living, and detect if things might be off. If an elder stops visiting the bathroom as frequently, for example, adult children can follow up on a possible urinary tract infection.
The danger, of course, is that elders feel surveilled by adult children or others.

Conversations around consent and best uses are critical with this kind of technology, particularly as more and more of these technologies are integrated into assisted living settings and other institutions.
A screenshot of People Power Family's homepage
people power family
People Power Family is an AI-powered system of motion detector sensors, which enables adult children to keep a proverbial eye on their aging parents from afar. The system, which has fall detection, also includes options for daily reports tracking when elders fall asleep, how often they move around, eat, and bathe etc. It is preferable to some other interventions which include cameras, considered more invasive by many elders.
The loneliness epidemic, particularly among elders, is well documented at this point; U.S. Surgeon General Dr. Vivek Murthy has called it a public health crisis. While human connection is, no doubt, the most profound cure for isolation, AI offers opportunities to create a sense of presence, information, and stimulation when loved ones can’t be around.
Used as a stop-gap and life enhancing tool, AI-powered robots are cutting down on loneliness and all the health impacts that it causes, allowing elders to age-in-place with more variety and vitality.
A screenshot of ElliQ's homepage
ElliQ is a proactive care companion, powered by AI, that includes an easy-to-use tablet, multi-user app, and continually contextualized prompts. For example, if a user just got a hip replacement, ElliQ will suggest seated stretches or introduce breath techniques, rather than checking in on the user’s typical walking routine. It also engages elders in mind-expanding conversations and multimedia experiences, like having coffee while visiting museum exhibitions from Paris. The robot, called a Time Best Invention of 2022, has shown the ability to decrease loneliness by 80%, according to the company.
What are individual caregivers doing with AI?
adult children
Adult children caring for aging parents are using ChatGPT to learn about different care options in their local area, such as in-home care, assisted living, or nursing homes. They can look more quickly at the pros and cons of each option, the costs involved, and how to choose the right option for their specific situation. They are also using AI to better understand particular health conditions and better plan for their aging parents’ future, such as making financial arrangements, creating a living will, or preparing for end-of-life care.
Parents are using ChatGPT in a wide variety of ways, including to learn about their child's developmental milestones, including physical, cognitive, and social-emotional development, addressing challenges like discipline, sibling rivalry, or sleep problems. Parents also find information on local parenting groups, educational programs, or financial assistance programs using AI. Parents are also using ChatGPT to plan for their family's future, such as saving for college, creating a family budget, or preparing for the transition to adulthood.
We had a great opportunity to talk to OpenAI team members Shyamal Anadkat and Glory Jain to ask them what they saw as the most important applications of ChatGPT in the care space. Here are some helpful considerations that Shyamal shared with us:
What should entrepreneurs think about as they integrate ChatGPT in their products/services?
“Heavily invest on customer experience.”
In other words, map AI’s best usage onto what you already know about your client’s deepest motivation for using the product/service and potential pain points. If you’re already doing human-centered design, it shouldn’t be too challenging to figure out where an AI boost could speed up your customer’s experience or more exquisitely target your customer’s needs.
“It’s going to be less about prompt engineering and more about dialogue engineering.”
In plain language, Shyamal is talking about the interactive and learning capabilities of AI. The care innovators who get the most out of AI will be those who think beyond the typical Google search, instead building experiences that allow customers to be more accurately and seamlessly served through more nuanced, value-driven, and specific content. These experiences don’t just give them search results, but put nuanced information to work in a client’s existing context.
“Ask yourself: How tolerant are you to some of the hallucinations in these models right now and what can you do when these hallucinations happen?”
We appreciate Shyamal’s caution here. These large language models are still in early stages, and as such, are not without errors. Entrepreneurs should think through risk mitigation strategies for AI-powered, client-facing experiences and not be naive about the potential consequences.
Here’s some of what we're reading, listening to, and watching regarding the future of AI:
“The A.I. Dilemma”
a video explainer with Tristan Harris and Asa Raskin, of the Center for Humane Technology, March 9, 2023
“Design and Artificial Intelligence”
a video explainer with Tristan Harris and Asa Raskin, of the Center for Humane Technology, March 9, 2023
"freaked out? we really can prepare for a.i."
a conversation between Ezra Klein and Kelsey Piper, both journalists who have researched A.I. for awhile and both parents
There is so much potential here to eradicate the grunt work of care, but like everyone else, we know there are real, valid fears around AI’s integration into the care system.
Our best bet, as we see it, is to keep asking ourselves: how does this technology facilitate or degrade human connection and thriving? How can we keep human beings, the people we love and care for, the main thing.

As Sal Khan said in his recent TED talk on AI, we must “Fight like hell for the positive use cases.” We will continue to ask the foundational questions as this technology continues to push us to the leading edge and beyond. We will continue to highlight and amplify the positive use cases in care. Get in touch with us and let us know how AI is changing your company's approach or your life as a caregiver!
Source: A national survey with n=2485 respondents to understand household willingness to pay for products and services that reduce the time spent on household management tasks. Conducted in April 2021 by McKinsey & Co.
Experiment with ChatGPT
Still struggling to wrap your head around what AI can do for care? You’re not alone. The best way to adjust to this new AI-assisted reality is to play around with the technology. We inserted the copy of this chapter into ChatGPT and asked it to write a 5-sentence summary. Here is what we got:
The article discusses the impact of Artificial Intelligence (AI) on the care system. The cognitive and logistical burden that falls on caregivers in the care system can be lifted with the help of AI. AI can also help in ensuring safety in caregiving and transform the aging process as Baby Boomers are turning 65. However, there are concerns about the unintended consequences of AI, just like with social media. The article highlights five areas where AI is already transforming care, such as care concierge, mental load, and caregiving. It also includes case studies of AI tools and platforms, such as Abridge and Milo, that help caregivers and patients in the care system.
Source: A national survey with n=2485 respondents to understand household willingness to pay for products and services that reduce the time spent on household management tasks. Conducted in April 2021 by McKinsey & Co.