#24 - Create the Customer is now The Healthcare Handout

New name, new format, same effortlessly cool author

If you’re looking for Create The Customer, you’re in the right place. CTC has become The Healthcare Handout, which is still a weekly newsletter, but with a new and improved format. I’ll be covering a larger number of things that happened this week in healthcare, serving you hot tips on which stories are worth reading, and providing some analysis along the way. I’m also introducing a new segment called Startup Of The Week which will be a brief glimpse into an interesting healthcare startup. Check it out ->

Hello from sunny, tropical Vermont, where most of our smarter residents have retreated to swimming holes. Like a weather reporter broadcasting from inside the buffeting winds of a hurricane, I’m enduring the heat and humidity to bring you the latest. Here are some updates from the wild world of health care.

Things That Happened

General Electric announced it will spin-off its healthcare business into a new publicly traded company which should be valued at around $60 - $70 billion, according to Bloomberg. The move is mutually beneficial for the healthcare division, which has seen fairly consistent growth over the past few years and a healthy operating margin, and the mother company which is trying to refocus the conglomerate and get out of debt, a manageable portion of which will be assigned to the healthcare entity.

“Even with the liability transfer, the health-care business should have flexibility to make acquisitions. That’s the most significant benefit from this portfolio shakeup” says Brooke Sutherland in Bloomberg. Get ready for some more integration of the vertical kind, GE Healthcare.

If you haven’t heard, Dr. Atul Gawande was named the CEO of the Amazon-Berkshire-JPMorgan healthcare venture. No, they still haven’t given it a name. And no, they haven’t said what they’ll actually be doing. Gawande, who is universally recognized within healthcare, was critiqued as an odd choice for having no real organizational experience. But STAT shows Gawande has done at least some leadership at Ariadne Labs where he served as Executive Director.

In addition to nabbing Gawande, Bezos and Amazon agreed to pay a reported $1 billion to acquire PillPack, a New Hampshire based mail-order pharmacy. Looking at both companies, and Amazon’s ambitions in healthcare, it seems PillPack was the “buy” option in their buy vs build analysis for a mail-order pharmacy. Included in the purchase price, Amazon gets mail-order pharmacy licensure for the entire U.S., relationships with drug wholesalers, pharmacy expertise, and a strong beachhead in the prescription drug business in the form of a mail-order pharmacy ready to be scaled. Here’s an interview with PillPack CEO TJ Parker from February with a bit more insight into what they’re up to.

Christina Farr reported that Walmart made a $700 million offer for PillPack earlier this year. You can fault Walmart for getting scooped, but it’s arguable that the relative value of PillPack and their licenses is higher to Amazon than to Walmart, who already has a thriving terrestrial pharmacy business. Had Walmart matched or exceeded the Amazon offer, it would have only been a defensive move to delay Amazon’s entry to the space via acquisition.

Things Worth Understanding

Things are not well with IBM’s Watson Health. While IBM put an early and friendly face on modern AI solutions, the self-selected standard-bearer is facing strong headwinds in finding a way to earn a profit on their more than $15 billion investment in Watson.

Make no mistake: Watson’s struggles are not an indictment of the future of AI in business. AI luminary Andrew Ng’s assertion that “AI is the new electricity” is not overestimating the effect of the tech on society. I prefer Benedict Evans’ comparison, where he likens the effect AI will have to the advent of relational databases. “Why relational databases? They were a new fundamental enabling layer that changed what computing could do.”

AI is promising, and thus the problems with Watson Health lie in the business model, or the lack thereof. When posed the question of Watson’s overall business model, IBM VP Ed Harbour answered “We have several business models for Watson.” Viewed from the angle of engineers at Phytel and Explorys, companies acquired into the Watson division, the several business models look more like indecision than opportunism: “IBM’s goal, the Phytel employees said, was to create a fancy new product that combined the capabilities of Phytel and Explorys. However, the offering managers didn’t have a clear idea of what that product would be. ‘They couldn’t decide on a road map,’ says the second engineer. ‘We pivoted so many times.’” These acquisitions, along with a third company Truven, were the hardest hit by the latest round of layoffs in the Watson division.

Competitive advantage in AI is derived from proprietary data and the quality of talent working on that data. The algorithms themselves are of lesser importance, hence why Google, for example, has open sourced ML frameworks in the form of TensorFlow. To create our own analogy, let’s say data is a field of grass, algorithms are cattle, and machine learning engineers are the farmers. The farmers can make better milk and beef by feeding their cattle on only the finest grasses, and by taking great care of the cattle with gentle massage and by reading them poetry. Assuming all calves are the same when born, the quality of the cow is determined primarily by the quality of the farmer, and the access to great, unique grasses.

Thus it’s concerning for IBM that they’re losing the ML talent battle, according to Jeffries in a recent exhaustive analysis of the division. These sought-after engineers are headed west towards the ML gold in San Francisco and Seattle, or rather the whopping paychecks and equity packages on offer at Alphabet, Amazon, and the many, many AI startups throughout the bay area.

Much of the AI future, it seems, will go to the vertical AI startup. This model, described by Bradford Cross, is typified by a small company that identifies a discrete problem, acquires or creates proprietary data around it, trains models on that data, and then delivers that core value directly using ML. Through this focus, these companies can attract mission-driven engineers, focus deeply on a single problem, and do so at a fraction of the cost that a behemoth like IBM might charge through consulting and setup fees.

As one Phytel engineer put it: ‘Phytel had more than 150 clients when it was acquired by IBM, the engineers say. Today it has about 80. ‘Smaller companies are eating us alive,’ says the first engineer. ‘They’re better, faster, cheaper. They’re winning our contracts, taking our customers, doing better at AI.’”

Things To Read

If you haven’t ever taken an afternoon to sit down and gain a foundational understanding of AI, machine learning, and what it all means for us, you’re past due. I suggest starting with Tim Urban’s overview in Wait But Why, and not just because I’m jealous of his drawing skills. For a deeper dive, the first half of Jeff Hawkins’ On Intelligence is a great resource.

Another day, another astronomical healthcare bill. Jenny Gold and Sarah Kliff highlight in Vox what can happen when an EMT determines a patient needs a trauma team ready when they arrive at the hospital. In this case, 3 hours of observation while the baby took a nap cost the family $18,000. The hospital’s response: “Zuckerberg San Francisco General Hospital spokesperson Brent Andrew defended the hospital’s fee of over $15,000, even though Jeong-whan didn’t require [trauma] services.”

“…signposts to guide you through the often bewildering landscape," is how Jorge Conde describes the 16+ Terms Entrepreneurs Should Know for Navigating the Healthcare Industry he defines in a blog post on the a16z website. I think “bewildering landscape” just might be my new favorite descriptor for the business of healthcare.

Startup Of The Week

Blueberry Pediatrics, part of the Y Combinator W2018 cohort, offers unlimited pediatric telemedicine for a $20 monthly fee. But wait, there’s more! Users get a kit with a thermometer, an otoscope camera that plugs into your phone, and a finger pulse oximeter, all for relaying biometric information to the physician. It’s basically Sherpaa for kids, but with gadgets.

It’s an interesting take on direct-to-consumer telemedicine, and one that has a potent appeal for parents. As one-half of the team responsible for ensuring my wobbly 16-month old son survives to adulthood, I can absolutely see it being worth $20 a month to make that frequent parenting decision of “does this warrant a trip to the doctor” a whole lot easier. Especially when $20/month covers all the kids in your family. At $240/year, if it saves your family one trip to urgent care it’s probably paid for itself.

Blueberry is targeting a major problem area in healthcare. If I take my child to the pediatrician and it’s not a wellness checkup, that’s a big bill that I’m responsible for. If my pediatrician is closed for the day, I’m headed to urgent care or the ER, which mean bigger and humongous bills respectively. Thus there’s a huge economic incentive for me to play it risky and hope my kid’s head bump wasn’t a concussion. And as any parent knows, that’s an excruciating decision. I’m interested to see the DTC uptake Blueberry finds - this is one to watch.

Things I Listened To

I wrote this week’s post to the expansive sounds of Alt-J’s An Awesome Wave (spotify), a fitting theme for any discussion of AI.


That’s all for this week. Enjoy your independence day celebrations, and I’ll see you next Sunday.