From Zuck’s office to personalized healthcare
Cofounder & CEO
It all started during the AI winter
I started to build and sell chatbots in 2002, when the AI winter scars of the 90’s made the mere mention of “AI” instantly destroy your VC pitch. Our bots helped businesses better serve their customers. Back then I was young and foolish, and I was convinced it would take no more than five years before bots fully replaced most human customer reps.
Fast-forward to March 2015. I’m sitting in Mark Zuckerberg’s “office” (actually a meeting room in the middle of the open space). Facebook had just acquired my second startup, Wit.ai, a developer platform that makes it easy to build apps that understand natural language. My ask to Zuck was bold and simple: can I hire a few hundred concierges to power my hybrid human+AI concierge service, Facebook M?
At this time, 13 years after I started my journey to build better bots, I had become wiser. I knew that some sophisticated tasks couldn’t be handled by AI alone: we should focus on AIs that can help humans do these tasks, instead of trying to replace them. With the immense powers of Facebook on my side – the best AI researchers in the world, around Yann LeCun, and practically unlimited financial resources – I decided that it was the right time and the right place to finally build this “augmenting” AI. Twenty-five minutes after our meeting started, Zuck gave me his green light.
We spent the next three years doing research on Facebook M. One of the most difficult and fascinating aspects of it was the interface between the AI and the concierges. There is a very fine line between an AI that bothers (hey, Clippy!) and one that is actually helpful. We learnt a lot, but M was never promoted to production. The key factor of this semi-failure was the very broad range of M’s perimeter; it’s really hard to be good in so many domains, with a loose definition of success (what ML engineers call “objective function”). The potential of the tech was incredible, though.
Taking the plunge into healthcare
That’s why, when I left Facebook in 2018, I started to look at several industries. My objective was to build a new “M” with a narrower domain, and have a big impact. Healthcare quickly appeared to be the most promising candidate – health systems are on the verge of collapsing everywhere – but also the most challenging one: there is a full cemetery of healthcare startups who “died by go-to-market”. We were scared to jump into this uncharted territory, even after hundreds of conversations with doctors and nurses who told us how much they needed what we were building. Then, one day, Yann “AI Godfather” LeCun came to our office and told us: “You guys should do it. Healthcare is the industry where AI will have the biggest positive impact. You are the right team to do that”. We jumped in and never looked back.
Back then, we didn’t know much about healthcare, but at least we knew that we didn’t know. In order to quickly learn from the field, we partnered with great doctors and nurses and became a medical provider ourselves. We started Nabla Health, a digital clinic for women in France and the UK. With our own physicians and patients, we discovered the real problems patients and healthcare professionals were facing – sometimes far from our initial engineer-anchored views.
I will never forget my immense frustration on April 8th, 2021, when we launched our digital clinic. We had hundreds of patient signups, and great press coverage. But at the end of the day, because of the medical nature of the business and the obvious privacy it requires, all the non-medical staff (including me!) had zero information on what happened. What problems our patients are trying to solve? Are they satisfied by their interactions with our medical team? Will they keep using the app? That day, I realized how healthcare was different from other industries. You have the same problems as others – acquisition, engagement, retention… – but you cannot use the same solutions. Hey, as a CEO I can’t even talk to one user!
From digital clinic to tech enabler
Day after day, we learned how to work with healthcare professionals – trying to improve their “customer experience” awareness and skills, without, obviously, ever crossing the medical line. We improved our healthcare delivery platform, step by step, so that it drives patient engagement and health outcomes. More and more doctors, clinics and startups started to ask us if they could use it with their own patients, from private practices to larger clinics, health startups, and huge institutions like AP-HP (the largest hospital group in Europe).
We are now ready to open our platform to them. Our first clients include several women health clinics in France and the UK (Aura Fertility, Omena, Jeen, Freya, Umaya…), the leader in digital oncology (Resilience), several remote patient monitoring and remote diagnostic clinics in the US (Cardiologs/Philips), and a virtual primary care provider in 9 African countries (Tchak). We are incredibly pumped to see, after several years of work, how our technology is being leveraged by our fellow healthcare providers for the benefits of their patients.
The journey is 1% done
As I write this, tech is crashing and Sequoia needs to publish their infamous RIP Good Times deck one more time. HealthTech is not immune to this downturn; on the contrary the healthcare bubble was inflated twice: first, like the rest of tech, by the free money markets, and second by the pandemics that forced so many patients into the arms of virtual providers. We see now that for many digital clinics, usage tends to plateau or even decline. A few of them are successful though; they found a way to provide exceptional, engaging care to their patients – often by blending asynchronous and synchronous care delivery. Generalizing this model to other clinics and other countries, and scaling it efficiently, remains a big challenge: this the exact problem Nabla is solving today.
Our mission is to enable personalized medicine at scale. Empathetic, predictive, proactive care will be provided by healthcare professionals who are fortified by AI-powered “super medical assistants”. The journey is 1% done.
Interested in this topic? Nabla is building a digital healthcare platform that enables personalized medicine at scale. Come join us!