- 3 minutes ago
TK
Category
😹
FunTranscript
00:05Hi, everyone. Joanna, thanks for being here with us today. You hit unicorn status this year.
00:13Congratulations. We did.
00:14And in the announcement, one of the first things you focused on in detail was AI. You're a women's
00:21health company, but you've prioritized that. Why was that important in your actual unicorn release?
00:28You know, I thought I was starting a menopause company, but it turns out I'm starting an AI
00:31company. And I have actually really learned that we can use AI in extraordinary ways to scale great
00:39care. So one of the questions we had when we first started the company is, how are you going to
00:43take
00:43care of millions of women and make sure they all get access to really high quality care? And I think
00:49there was rightly some skepticism there. How am I going to train thousands of providers to make sure
00:54they all follow care protocols? How do I make sure that we have standardization across the country
00:59as we scale? So it turns out that AI is magnificent for that. And what we've found is that we
01:04can use
01:05AI to train providers, to create really good care protocols, to ensure quality, plus for efficiency
01:13in a way that is making our company much more honestly exciting and higher quality care than I
01:19ever could have hoped before we were using AI. So tell us how you get to that higher quality care.
01:24I think you've created your own LLM effectively. So it's a RAG, which is Retrieval Augmented Generation,
01:31some, I'm going to get those words wrong, but the point is there is open AI, okay? Open AI pulls
01:39from the
01:39whole internet. And that's a lot of old outdated data with regard to women. So before 1992, women were not
01:47even
01:47included in clinical studies. There's also a really bad study done in 2002 that really got women's health
01:55wrong. And for many, for 20 years, women got the wrong health information because of that. What does that mean?
02:02It means all the data in those 20 years published was wrong and misinterpreted. So if you have an open
02:08AI,
02:09if you have an open system, it pulls from those studies and gives women incorrect data. So we had to
02:15start over and we had to
02:16create our own care protocols and we engage doctors from around the world to actually help us create care
02:22protocols. And then we have turned that into a system that all of our providers can go and search to
02:29make
02:29sure that they give the right care to women. So it's actually really rewarding, I have to tell you, the
02:36nurse practitioners who are providing the care are just astounded by how good providers they are now
02:41because they have access to all this data. So they're able to put in very complex cases and get
02:47really good information back on what's working for the patients, what did work using all of our
02:52historical information as well. So we're really able to scale really good care. And there's a lot of
02:57skepticism in the public toward AI and toward about hallucinations and its reliability. So how, how,
03:06tell us about working with these nurse practitioners. How did you get them to buy into it and how have
03:11they become convinced? Yeah. So honestly, AI transformation in a company is hard. You have
03:18a lot of work you need to do to get people to buy in. So we had an entire process
03:23where we started with,
03:24you know, about 10, 10 of them and they were testing it. And it was like, can you beat the
03:29AI? Can you,
03:30are you going to trick it? And they did find things that it wasn't doing right, report back. And then
03:35we iterated,
03:35we added 20 more and then we iterated and then we slowly rolled it out to the whole company.
03:40And so a lot of it, when you're thinking about AI transformation in a company, a lot of it is
03:45getting people to buy in and getting them to feel comfortable and getting them to, to trust the
03:49technology. And it took us a while to get there, but now we have hundreds of them using it.
03:54And what was like, can you give us an example of a turning point where you like change someone's
03:59mind? Like you telling me how you went around the country talking to people doing this?
04:03So that's just not with our NPs. What I've been doing is actually going around the country and
04:08having dinner with our employees because the most, the company is mostly remote. And I asked them,
04:12how are you using AI in your job? And, uh, often they, they're not right. And then I would say,
04:18okay, so what are you working on today? And a woman said, well, I have this assignment. I have to
04:23go
04:23review 150 contracts for, to see whether or not we needed to use a translation service in certain states,
04:30let's say. And she said it was going to take her a month to review all these contracts. I said,
04:34oh,
04:35well, let's go try Gemini. And we pulled up her laptop at dinner and I showed her how to put
04:40all
04:40the contracts into Gemini and then asked Gemini the questions. And it took her 10 minutes.
04:47And then you see her eye light up and she's like, oh my goodness, like you just made my life
04:52so much
04:52easier. And, uh, so that was, when that first happened, I was like, okay, I'm going to go have
04:57dinner with everyone around the country. So now I go have these dinners and we host the dinners and
05:03we sit down and we say, okay, what are the things you're learning? And then we've also actually opened
05:07AI office hours. So we have engineers who we recruit to come and answer the questions from the
05:12non-engineering team and basically help different people on the team to AI-ify themselves. And, um,
05:18I think of it a little bit like Henry Ford with an AI toolbox. Like how do we basically go
05:23around
05:23the company, be Henry Ford, make automation, um, but do it in a way that people are comfortable
05:28with. And honestly, for us, they're not losing their jobs. It's not threatening. It really is about
05:33augmenting their jobs and making them easier and making their jobs, honestly, far more interesting.
05:39And does it give you any like insights into how AI might or might not be effective in healthcare in
05:45terms of truly getting the better outcomes? Because that's one thing that has been talked
05:49about a lot, but there's a little skepticism, like, when is this going to happen?
05:53Well, I think already it's getting better outcomes. So I'm not, I'm not skeptical about that. I think
05:59that if you think about going to one of your, if you go to your local primary care physician,
06:05the likelihood of them knowing every single question that you have is just really,
06:10it's hard. There's just too much data out there. And we're, we as humans are complicated beings,
06:15right? We have a lot of questions. What supplements should I take? How much of this one food should
06:19I take? How much should I, you know, there's so many questions. You can't expect that doctor to know
06:23it all. And then they leave and they come in there. So the women are so frustrated. Well,
06:26my doctor didn't listen to me. Well, the doctor had 15 minutes and had to answer a lot of questions.
06:32So I just feel like the AI, when you use it appropriately, and you know, when you are able to
06:38get the
06:38right information out there, you're able to be a much, much better provider using AI. And so
06:43most of our providers just feel very empowered by the fact that they have access to this really
06:48good information. And do you feel like happy where you are? I think you mentioned like you wish you'd
06:53done it even sooner. Like what's the time? How are you feeling where you are in terms of your path
06:58on
06:58this? I mean, I suppose the challenge is the technology is changing so fast. So every few months,
07:06we have to look and say, are we up to date on what we're doing? And make sure that we
07:11are,
07:12and then make sure the whole company is following along. And you know, that's a challenge. I mean,
07:16operationally, bringing AI to your entire company and getting everyone to change how they are doing
07:21things is a challenge. But the more comfortable you get the entire company with the technology,
07:27the more they're willing to explore and use it with you. I mean, I'll tell you this morning,
07:31I got to work early. And the head of my billing department was there. And I was like, Chris,
07:35how are you doing with AI today? And his eyes lit up. And he was so excited to spend the
07:41next 20
07:41minutes telling me about the different software that he was implementing and that he was making
07:46his own code. I mean, and I said, you didn't really expect, you know, even a year ago, you'd be
07:50doing
07:50this. And he didn't. So it's actually really fun when you see that someone kind of wakes up and
07:55realizes all the things they can actually do, even when they're not an engineer.
07:59That whole vibe coding phenomenon. And we'll see what it means for the software companies.
08:04So some of your investors are very famous celebrities, Amy Schumer, Connie Britton,
08:11Tory Burch. And was that intentional? Like, do you think that's important part of your message to
08:16have some pretty, like, impressive investors well known? Have you noticed that the media likes to
08:22write about celebrities? There you go. So that was intentional, yeah.
08:28The media really likes to write about celebrities. It's just crazy. I mean,
08:33I also have some really impressive funds as investors, but everyone wants to talk about the
08:38angel investors who put in, who are celebrities, who put in money. It's just true that people like
08:45to talk about celebrities. So it's a good strategy to have them invest in your company because people
08:50pay more attention to it. The other thing that I found is that, honestly, people often trust
08:56influencers online far more than they trust doctors. And so micro influencers maybe have,
09:03you know, 10, 15,000 followers sometimes, and their followers will absolutely trust them more
09:08than they will trust their position. And so we are, you know, we have to only work with credible
09:15Instagram celebrities or, you know, Instagram names. But people really do trust celebrities. And so
09:21getting them on board has actually been a very successful strategy for us.
09:25So that's part of what you do is go out and find brand ambassadors for you online?
09:30We do. As long as they follow, I mean, they have to really follow our care protocols. We're very
09:35choosy about who we work with. But yes, we do find that actually working with Instagram influencers
09:41and celebrities is a really good strategy to get your name out.
09:44There you go. So, excuse me. Another very interesting thing in terms of investors and fundraising,
09:55you've mentioned, you got a lot of no's. I mean, you're a unicorn and you've raised a lot of money,
10:01but you got a lot of no's along the way. So how do you process something like that? And what
10:07do you do
10:07with that information? I think the best thing I learned is that no's are actually only not now.
10:15And I know it's a pat thing, but it really is true. And it is certainly true in fundraising.
10:20So there were a number of people who I approached early on. Honestly, menopause has become a really
10:25hot word, right? But a few years ago, menopause was not hot. And it felt hot to me, but it
10:32was not
10:33hot for everyone else that I was talking to. And I was having a really hard time getting people to
10:39pay attention to me. And I would go and I would give them the argument and explain to them why
10:42they
10:43should matter to them. And they would turn me down. But by the way, you already had a successful exit.
10:47I mean, you had like all these connections. It was still really hard to convince people this was a market.
10:52I mean, women's health has exploded recently, but it was not a market a few years ago. I can't.
10:57You explained once that the VC needs to see a potential, what, $10 billion potential, right?
11:06Yeah. They have to really believe it can be a big company. And explaining to them why it could be
11:10a
11:10big company is hard. So many people say no. They're just always going to say no. And I just decided
11:16that
11:16I would just stay in touch with them after they say no and send them updates and go on walks
11:21with them
11:21and just stay friends and keep going. And honestly, some of the firms that turned me down two or three
11:28times ended up investing later. So it was definitely the case. I mean, some of them still said no forever.
11:34But a lot of the ones that I really stayed friends with, I was able to bring them along to
11:42a yes eventually.
11:44And the no's, what did you do with the information? I think you talked it over with your team.
11:48Like you, can you talk a little about that? Yeah, I used to take notes. Now I would do it
11:52on
11:53AI, but now I, before I had to write it or type it. But I used to just go back
11:59and tell everyone,
12:00here's what they said. Here's what they didn't like. Here's what they liked. Here's what the feedback was.
12:04Do we like this feedback? Do we think it's relevant? Let's take what we can from it and
12:08ignore the rest. But honestly, a lot of them were pretty smart in their feedback. And so I thought it
12:12was
12:13helpful to share. That's great. I'm going to go to questions in a minute if anyone has any. But you,
12:22you've had a long career before you got to this job, to this role, which you started like five years
12:29ago.
12:31And yet, you know, we see Karina, like there's people are dropping out of college, high school to
12:37start companies. So do you have a view like, no, it's better to wait? I mean, you got had so
12:43much
12:43experience and connections or like, do you feel like now's a good time in your life to be running
12:50a company? Like, can you give us a sense of the whole age and founder thing? Well, I saw her
12:55and I was like,
12:56wow, I did not have that level of confidence in myself when I was 24. Yeah. So I have a
13:01lot of
13:01admiration for that. You know, I think it really depends on where you are in your life. For me,
13:07I'm so lucky my kids are gone. That is the best. Like, I have so much time. I can't get
13:15over how
13:16much time I have. And honestly, I have more energy. I have time. I don't have to cook dinner for
13:22anyone.
13:23I don't even have to think about cooking dinner for anyone. So for me, starting a company now my kids
13:28are gone is really great. But obviously, for everyone, it's whenever I think it's the right
13:35time for them in their own life. But I feel really incredibly lucky that I get to do this in
13:40the
13:40second half of my life. So I just feel very fortunate. That's great. And you probably have
13:44some great health care. Are there any questions in the room? Anybody? You may just let us know who you
13:55are. And I think I know who you are. This is Alistair from Business Insider.
14:01Like, you have an AI-powered product. And I wondered whether you worry about
14:08being a kind of AI wrapper. So the idea you mentioned open AI, and it answers health questions,
14:13and then there's open evidence, and then Google answers health questions. Do you feel like there's a
14:20sustainable place for, you know, vertical specialists in this area?
14:26So we at MITI are insurance-covered care. I don't see a world at any time soon where
14:35insurance-covered care is provided by AI. And I also think that despite a few unique things happening,
14:44that most care, most prescribing will be done by providers. Although one state is doing a test
14:54around this, in general, doctors and the state medical institutions are not so fast to let AI do
15:02prescribing. So what I feel really good about is we're able to offer insurance-covered care in all 50
15:08states. I would love you to come eventually search my AI, do all the research before you come to the
15:14provider, and then say, okay, clinician, do you agree with this AI? Is this the right thing? And we can
15:21discuss what I just learned from my AI search, and then together you'll decide whether that's the right
15:27approach to your care. But I don't think that doctors and clinicians are going to be replaced. I just think
15:33it's very much around augmentation. Because what you're describing is probably happening. I think
15:39most of us go to AI first now, right? Right. That's exactly right. And the problem is if you go
15:44to the
15:44open AI, then people come to us a lot with the wrong information. So we spend a lot of time
15:50debunking the things they found, they saw on AI because it doesn't include only the most updated
15:54women's health research. So, you know, I'm hoping over time we'll actually be able to, our AI will be good
16:00enough that we can expose it to the public, and then people can search it themselves before their
16:04visit. Oh, you're specific. Yeah. Yeah. Okay. Right. So what I'd like to do is create my own vertical
16:09AI for women's health, and then open that to other people. We're not, we're not there yet. But that's
16:14certainly a goal that we have. And there's one question in the back here. Just let us know who you
16:22are.
16:25Hi, Samit from Crafting. You mentioned that you want to make it more vertical, but the way AI
16:31agents are integrating with each other is evolving so rapidly. And we're also seeing people interact
16:35with it in new and novel ways on a pretty regular basis. How do you think the average user of
16:41MIDI
16:41might actually change how they interact with AI over the next year, whether it's a nurse practitioner,
16:46a patient or somebody else, maybe six to 12 months from now? You know, honestly, I'm making a bet on
16:53the fact that people still want to talk to someone for their care. Other companies are probably going
16:59to make a different bet. But we have found that what we are going to be really good at is
17:03empathy,
17:04and that empathetic person who understands you, who asks you a bunch of questions and really gets to
17:10know you. And then we're offering longitudinal care over a long period of time with the same provider.
17:14So it's not just episodic care. I still believe people want that. So from my vantage point,
17:20I'm going to use AI to make the care better, to make the tools better for efficiency purposes,
17:26and to ensure high quality. But the bet I'm still making is on the human interaction. And so I don't
17:33see and I don't see that going away. I mean, it's a joke at our company that when we first
17:37started,
17:38people would come in every day and say, you know, thank you for the care. But what they said was,
17:41I felt seen and heard and cared for. And our engineers used to read this every day,
17:48and they're like, wait, are we a medical company? Like, why is it that people are thanking us so
17:52much for feeling seen and heard and cared for? And I think that's still what people want. So
17:57that is certainly the bet that we're making.
18:00Isn't that answer? Yes, you are? When the engineers asked you that?
18:06Well, that's what I said. Well, so when they asked that, I'm saying, yes, we're doing the medical care.
18:10And what people are really missing is actually the emotional connection of someone who cares
18:15about them. And that's what we're doing. And that's what we get thanked for more than everyone
18:18else. Like, it's the emotional connection. I don't think that's going away. I don't think,
18:22we don't have a sense that people really want to get their care from AI. We get the sense they
18:27want
18:27to do research, and then they want to talk to someone about their research. But I don't know,
18:33we so far, our business is not suffering from people doing a lot of AI searches.
18:39Emotional connection. A very close friend of yours, Susan Wojcicki, passed away a couple of
18:46years ago, one of the great leaders of Silicon Valley, YouTube. How does her legacy play a role
18:53in your leadership now? So, yes, it's true. Susan was my oldest friend, and we were friends since we
19:00were 10. And I had the opportunity to watch Google get started in her house. And I used to go
19:05over there
19:06all the time. But to me, she was just a normal person. Like, she wasn't a luminary. And it's
19:12actually incredibly helpful in your life to see normal people do great things. I find that,
19:17for me, it's very inspiring. You know, and then, I mean, so watching her grow to build this huge
19:24company and still be a very normal human being is actually both reassuring to me, but also it shows
19:30that you don't have to be a super person to do these things. You just have to actually be brave
19:34and
19:35decide to go do it. Thank you so much. Great to have you, Johanna. Thank you.
Comments