"Why can't it be both?"
An interview with the founder of Care.com
The idea that the U.S. childcare system is fundamentally flawed, if not totally broken, is commonplace in the circles I run in. Until quite recently, this was an abstract problem for me. Now, as I confront years-long daycare waitlists and monthly costs that rival my mortgage payments, I feel that brokenness quite viscerally.
To fix our care system, we’re going to need widespread cultural and policy change. More of us need to acknowledge that raising the next generation is a social rather than individual responsibility, for example. Lawmakers need to step up and provide subsidies or other mechanisms that make care more affordable for families and more sustainable for care providers.
I’ll continue to advocate for these changes. Without them, high-quality, (relatively) minimal-stress care is likely to remain out of reach to all but the upper-middle-class. And yet, I’m a pragmatist at my core. I believe that alongside the activists and social change-makers, we need entrepreneurs and businesspeople solving for specific pain points and marshaling capitalism’s might for good rather than ill.
Sheila Lirio Marcelo, my interview guest this week, has made a career of doing just that. If you’ve ever searched for a babysitter or nanny, you’ve likely come across her work: she founded Care.com, one of the largest digital marketplaces for finding childcare (or eldercare, or tutoring services), back in the early 2000s. She stepped down as CEO in 2019 and has more recently turned her attention to building Ohai.ai, an AI-powered assistant designed to reduce caregivers’ cognitive load.
As long-time readers know, I’m on-record as a bit of a skeptic (see the linked post below) when it comes to using AI and other digital tools to reduce or redistribute cognitive labor. As you’ll read, I admitted as much to Marcelo before asking her to make a more optimistic case. And to my surprise, she changed my mind. Not fully; I still think there are a lot of limitations. But I was impressed by Marcelo’s insights about human behavior and her more gradualist approach to integrating tech into family life. If anyone can crack this particular nut, it may well be her.
Read on for a fascinating discussion about how Match.com convinced Marcelo the world was ready to find care online; why she has gradually broadened her definition of “care”; and how learning to work with a digital assistant is a little like learning to work with a human assistant. Our conversation has been condensed and lightly edited for clarity.1
If you’d like to support more in-depth interviews and analysis like this in 2025, consider subscribing or sharing this post with a friend!
AD: When you first started working on issues related to care 20-ish years ago, what were the big problems or pain points that you were identifying and trying to solve?
SLM: I had written a business plan when I was in grad school, and it was an online marketplace for real estate. And then I got pregnant with my second child in grad school, and my professors were kind enough to say, ‘Hey, come teach here.’…I was grading internet business plans [as part of my teaching job], but I felt like, you know, I probably should learn how to operate these businesses. That was what really led me to join a startup. And so I joined Upromise, learned the ropes, did a bunch of different functional jobs.
But then what led me back to care was really my own personal situation. I had two kids, my parents were helping care for our second one, and then my father ended up falling ill. And so here I was, sandwiched [between dependent children and parents]. At the time, the dynamics in the market were that the Yellow Pages were going away, classifieds were going away. I was an immigrant in the US, and didn't have a lot of support. And so I was like, Where was I going to go?
I was like, ‘Wait, how about a marketplace for care?’ And so that's really how it started. I think it was just a proximity to the issue and the challenges that we were facing. There just weren't a lot of options that were also affordable. Because at the time, I'd just gotten out of grad school, I was working at a startup. My husband was in social work, and so I couldn't afford an agency. I knew that it was going to be me posting something and trying to just do it myself. I started to use Craigslist, didn't feel safe, didn't see a lot of options. And so I was immersed in the actual experience of trying to find care for both my kids and my parents. It was just difficult.
And then the other thing I observed is that we were starting to feel comfortable with people-to-people services online. Because Monster.com [the job board] was I think five years old at the time, and so we were starting to feel comfortable to find people for jobs online. And then Match.com. So I said, maybe people will be ready to find care online. Those two companies set the tone for people-to-people services, and that mattered a lot. Good friends of mine had tried to start a care company 10 years before, but I don't think the market mindset was ready. And then the other thing was, social media was starting, and people were starting to post their personal lives online. And so I think there was just a comfort in the marketplace that I observed, and then I just had a personal passion to solve things.
AD: There’s this idea that care for one’s children or one’s parents is one of the most intimate kinds of services that can be provided. So needing that proof of, ‘Okay, people are trusting the internet’—I hadn’t thought about that angle, but it makes a lot of sense.
SLM: What was interesting to me is care was often described as sort of personal responsibility and love. And anyone who talked in capitalist economic terms separated it. And even now, I find the two worlds are always separated. And I'm like, why can't they be both described in the same way? Why can't care be described as driving the economy? Really, I mean, people can't work if they don’t have care for their loved ones, and at the same time, it is a responsibility that we have. Somehow we tend to label it in a way that just says it's one thing. And I do a lot of advocacy to say, why can't it be both?
AD: Although you’re still working in the care industry, you exited Care.com a few years ago. When you were leaving, what was your assessment of, ‘okay, we’ve made progress on X, Y, Z, but here are the things that are still not working in the care system’?
SLM: I had a wrong thesis. I thought senior care was going to take off when we launched Care.com in 2006, but it took us about 10 years. 2016 was when we started to see it really develop. I was thrilled that senior care was also catching up [to childcare] on the marketplace side. Affordability of care was still something that I cared deeply about. It hadn't been solved.
And then the other big thing for me was what I call “everyday life coordination,” which is a big part of care and part of the reason we launched housekeeping and tutoring [as services you could search for on care.com]. People were like, ‘Why did you launch tutoring?’ I define that as care. Even though, again, we try to segment these things from a business standpoint, versus what is the overall role that they play at home. And it's the same with housekeeping. Like, if I didn't do laundry, that would impact how comfortable and confident my kids feel in school. And so I kind of define care broadly in these roles.
Coordination was another one of those. What we quickly realized when I was running Care.com is that families struggle with coordination. Like, that is its own category of care. We fulfilled services, but we didn't fulfill the majority of care that really happened beyond a paid caregiver coming. There was this whole other world that we knew about, and there were really no solutions for that. And so this new company, I say it's like, the Care 2.0. I'm still in the care industry.
AD: You're very optimistic, I imagine, about, how AI and digital technology can really solve some of this care coordination that we've been talking about. And I'll admit that I have looked at some tools in this space and been a little skeptical. Like, what's the added overhead of managing the tools? And, is the technology good enough to get to know my family's priorities and values? But I'm amenable to having my mind changed. I know that this space is advancing incredibly rapidly. So I’d love to hear the optimistic case for how this is actually going to make a dent in these problems we've been talking about.
SLM: I'm learning a lot. So last year, we were developing a generative AI course to complete tasks for developers. And I said to my team, ‘Is the technology now ready to complete tasks for families?’ My team got really excited. We interviewed 200 families, and they kind of described all their different needs and where they could use the help. We were like, wow, there could be something here.
I have had an assistant for 20 years as a chief executive, and so I've had the fortunate help of incredible women in my life. And so I really thought through, like, how do we [introduce the technology] in a way where you build trust? You also have to educate people, because not everyone's had an assistant, that’s a small sliver of a percentage of people. So how do you roll this out in a way that people can even understand how to delegate? There has to be—just like I was telling you about a norm change with Monster and Match. I'm not blind to the fact that, okay, we are in a new AI world, but there's a behavior change of delegating things you normally wouldn't delegate.
And so I really thought through it. I was like, how did I ease into learning how to be productive with an assistant’s help? There were basic things that we really thought about, like, what are the highest impact, highest volume things that can create a sense that this thing is working?
For us, the number one thing was a household calendar. One of the dynamics that we heard is that many families use a digital calendar, but there's not huge compliance from the partner. [i.e., one partner uses the digital calendar intensively, the other doesn’t.] And we added other features, like reminders and to-do lists.
Then users told us, ‘Oh my gosh, I get an 11-page newsletter from school. And I often miss the message, and I feel like a terrible mom.’ And so we were like, okay, email. And schools still communicate with paper-based documents. They come home with the kids, and then you're like, ‘Well, what do I do with this?’ And you have to sit and create a pile, and you hopefully get to it later in the evening. So what we've done is, hey, we'll take a picture, we'll summarize the event, and we'll put it on the calendar.
The way we view it is, we're taking a step at a time, working with our users, making sure that every release and feature is really effective. We don't release a feature unless 90% of the task is actually completed by the AI. So it really is accomplishing it for the family. We're taking a very methodical approach around, what do users want that they're telling us they need help on, and what's also practical? What are you willing to delegate?
I mean, we're really an assistant. We're not solving, we are supporting and enhancing based on what works for a household, because every household is different, which makes what we're building very complicated. We have to build it with the flexibility of, how does it work? And the chief household officer varies by household. I don't assume that it's always a woman, because it certainly wasn't in mine.
AD: One thing I'm curious about, and you alluded to this when you were talking about calendar management, is that I've heard from people, mostly women, that when they want to adopt a new system, it’s hard getting the partner to buy in and making it something that is not just another system that the chief household officer has to manage. So what does that look like? How are you thinking about that issue?
SLM: We really researched it. We call it “BYOC,” bring your own calendar. We're not replacing your calendar. If you started a digital calendar and your issue is adherence to it, we're not going to go pitch you another calendar, because that doesn't solve your problem. And so we get their existing calendar, and then we send reminders. ‘We notice that you're double booked.’ We also we keep sending the reminder until the task is done or the conflict is resolved.
That's how we do the compliance. It's not that hubby or partner has to actually put in the event. We do it for them. So versus this feeling that, ‘Oh, my God, I now I've got this huge burden, I’ve got to work on this calendar that Sheila assigned me.’ No, the whole point is the assistant's the one guiding Ron, my husband, to say, ‘Have you done this? Would you like me to update it?’
AD: One of the things that I hear about in my own interviews with families, and from women in particular, is ‘I don't want to nag.’ And it sort of sounds like the assistant is in the position of offering those gentle nudges and gentle reminders, so the partner doesn't have to do it.
SLM: And the gentle nudges are on both sides, so that there is a shared accountability that happens around coordination, and there's also visibility around it. Versus this feeling like I have to hold it all on my own and I don't understand why I have to be the one.
AD: As a final question, if you think about a year from now, maybe five years from now, where do you hope the family technology space will be?
SLM: We're going to lean more and more on digital help. What I worry about is the same unintended consequences that happened on social media or that happened with the iPad. Where I worry, you know, and one of the reasons I'm building again, is how do we use the power of technology to assist and not replace?
I'm worried that we start using AI to read a book to a child going to bed. Do I believe that will happen? I do, actually, which is sad. I'm very much invested in human connection. I do it in my leadership. I do it as a mother. I do it as a daughter. I do it now as a grandmother. And so I'm constantly thinking about, how do we develop responsible technology to actually enhance our connection with people, versus replacing that connection? Unless leaders are conscious about it, and we're building things with that in mind, I worry about that a little bit. And I'm not trying to say, ‘Oh, I'm freeing up your time so you can do more work.’ I'm hoping that it's about enhancing human connection, is really what I'm focused on.
You can learn more about Sheila Lirio Marcelo here, and about Ohai.ai here.
One thing you learn quickly as an interviewer is that humans tend to communicate a lot differently in writing than they do in speech. Perhaps I should just start a podcast, but so long as I publish written interviews, my goal is always to preserve the speaker’s meaning as accurately as possible while making the reading experience pleasant for all of you.