Transcarent, AI Therapists, Interoperability, and more

This episode of Out-Of-Pocket is brought to you by…

Healthcare compliance, simplified. Vanta automates SOC 2, HIPAA, and HITRUST CSF compliance, empowering health technology companies to protect sensitive data and demonstrate trust.

Download Vanta’s SOC 2 checklist to learn more about the SOC 2 compliance process and the road ahead.

Want to be featured in the newsletter? Check out OOP’s sponsorship options.

—

The JPMorgan Healthcare Effect

I hope you all enjoyed the JPM healthcare conference in San Francisco. I am not there because I am happy, but I always found the event so fun because of all the healthcare ironies it embodies.

1. Most people go there but can’t actually go to the best parts (like the main conference itself). Almost like access is a challenge…
2. You awkwardly avoid the open air mental health crisis happening in SF as you go to a party sponsored by a mental health startup.
3. You 100% will leave sicker than you got there, conference-acquired infection (CAI) is a quality measure.
4. Everyone marks up the prices just because they know it’s healthcare (which you don’t care, your employer is covering it).
5. All the meetings are measured in increments of 15 minutes.

All love - I actually always liked the energy of JPM and thought it was a great way to create a circadian rhythm in the industry. Many companies push to get things done so they can announce it at JPMorgan, and it’s one of the only times it feels like companies have a sense of urgency in healthcare.

Give me any good tea you learned while you were there. In the meantime, here are a bunch of micro vignettes as a substitute for TikTok getting banned.

Solo founders

Anecdotally - I’ve been seeing way more solo founders starting digital health companies nowadays. And lots of them are choosing to avoid the VC route completely. 

My guess is that this is due to a few things:

1. Co-founder breakups are a very common way for a company to die or a key person to get pushed out of a company. As more people have started companies, more founders end up experiencing this. Solo founding makes people feel more in control of their leadership.
2. Founders want to own more the equity pool so that if they get diluted as the company raises money or want to sell secondary shares, they still retain more of the company.
3. A whole host of agencies dedicated to healthcare + no-code/low code tools have made it much easier to get a working prototype together without a technical co-founder.
4. More founders are thinking about bootstrapping, which is more viable with only one mouth to feed.

I think there are pros and cons to going solo. Having someone who can weather the ups and downs of a company with you can really help the founder journey, which is a pretty lonely job. It can also be helpful to have another peer you deeply trust that can handle different parts of the business, provide different perspectives with the same level of emotional investment and context, etc. 

But maybe there’s some merit to going solo? You can probably be more uncompromising in what your vision for the company is. You could also probably sell the company for a smaller amount that would be a much more meaningful windfall for a solo founder.

Maybe you can do more interesting things with your company equity if you own more. You could give more equity to the early team to attract really good talent early on. You could build a larger network of advisors or have first customers have more meaningful equity in the company. 

I’d be curious what the long-term success rate of solo founders looks like, and if the prevalence is actually increasing or are lonely people just more attracted to me.

Hacking together interoperability

A few loose thoughts on how AI in healthcare is hacking together interoperability:

1) Generative AI might actually achieve the future that robotic process automation promised. The issue with v1 RPA was that because it was focused on screen specific features it broke frequently + was too difficult for frontline staff to learn the tools to make fixes themselves.  

GenAI tools look at intent on the page instead of where the specific button is + the interface is getting much easier to use to make fixes (e.g. just use a chatbox). 

One version of this is AI agents that are scraping data out of EHRs on behalf of providers. There are some very big lawsuits being fought around this exact idea - if a doctor customer authorizes a bot to go into the product and pull data out, should the EHR be able to stop the bot? (Intus v. RTZ, PointClickCare v. Real Time Medical Systems).

You can see some product examples of this with NexHealth Synchronizer and Solv’s RPA tool which explicitly try to work on EHRs that won’t play nice.

2) Voice AI can create information flows where APIs don't currently exist today. Pharmacies don't have APIs that let you see their inventory, but you can call into a pharmacy to ask if they have something in stock. It’s hard to find the next available appointment for a doctor, but you can call the front desk to ask.  You can do this to essentially create ad-hoc data pulls.

3) We are seeing companies that take advantage of ubiquitous data transportation standards (e.g. fax) and expanding the use cases by being able to do read anything that comes through, regardless of how the data is structured (e.g. see Tennr as an example).

It’s possible that instead of trying to regulate our way into interoperability, AI tools will actually bring that future forward by force.

Transcarent-Accolade and At-Risk Fees

There are only three known business models in digital health.

1. D2C - Direct to consumer
2. B2B - Business to Business
3. GT2E - Glen Tullman to Employers

As many of you probably saw, Transcarent acquired Accolade for $621M. Accolade essentially helps employees navigate their benefits, figure out where to get care/drugs, offer virtual care/second opinions, etc. Transcarent does something similar but instead of a call center telling me my drug is not covered, it’s an AI bot that’s telling me. Innovation is beautiful.

This acquisition is an embodiment of the next phase of the employer healthcare market. Phase 1 was companies like Castlight just trying to organize information so employees could make more informed decisions. Phase 2 was companies like Accolade that were trying to provide employees with some level of triaging + guidance. And Phase 3 are companies doing all of the above and also attaching some financial outcomes if they’re successful. 

Transcarent adds the layer of enabling contracting between employers and providers and attaching ways they can get paid for driving some of those results (e.g. steering employees to surgery centers with pre-negotiated bundled rates). This contracting piece for self-insured employers has usually been the domain of health insurance carriers, but Transcarent is trying to add that to their navigation features.

More care delivery companies in the employer space are pushing their fees at risk. But now we’re starting to see all-in-one care navigators trying to go broader - Included Health, Blue Shield, and CalPERS recent deal is another example of that. 

“A key component of the new five-year contracts is the establishment of performance guarantees in which Blue Shield and Included Health will put $464 million at risk if they don't meet the program's goals for controlling medical cost trends and improving quality.

The contracts set the initial medical trend cost target at 5.5% in 2025 and lowers the target each year until it reaches 3% by 2029. If CalPERS' trend is lower than the target, Blue Shield and Included Health stand to share in the savings.” - Healthcare Finance

I think this is a trend we’re going to continue to see, especially as employers feel more pressure as fiduciaries and are pissed at their current network contracting partners who are getting them shit deals. My guess is that you’ll start seeing way more consolidation in the next 6 months as more we-do-it-all navigation players try to enter these “fees at-risk” engagement. 

For employers, the push for fees tied to outcomes seems to have started with specialized providers e.g. A1c management and weight loss for diabetes management or surgery avoidance for musculoskeletal companies. But how will these specialty specific at-risk service providers interact with companies like Transcarent offering the “all-in-one” package? 

Which brings me to my next thought…

Interlude

SIIKKKEE first I have to shill some stuff, bringing you salvation from the hell of trying to figure out all this healthcare stuff yourself.

We have a few crash courses that are starting very soon! This is only for people that want to get really smart on healthcare quickly, not people that like leisurely learning on a stroll.

• US Healthcare 101 (starts 1/28, enrollment ends soon) - Taught by yours truly that goes through the basic money flows, major stakeholders/laws, and things happening in US healthcare
• EHR data 101 (starts 1/28, enrollment ends soon) - You think just because you have access to EHR data your pain is over?  hahahaha. we’ll teach you everything about formats, how to run analyses properly, etc.
• How to Sell to Health Systems (starts 2/3) - If you’re selling to hospitals, you need a prayer and a playbook. This course will give you the latter.
• Healthcare Product 201 (starts 2/18, this is the final time we’re doing it) - Everything you need to know about how to bring a healthcare product to market, build a business case, etc.

Wanna chat about group rates or bundles? Hit me up, we can talk.

The Point Solution Issue

A lot of the commentary in the last few years has been around how no one wants to buy point solutions any more. This has led to lots of rollups and M&A of various point solutions into larger companies.

While I don’t disagree that this is true, I think this is also bad!

You know the tiny hill I’d die on? Point solutions are a sign of a functioning market and we should want more of them. Point solutions try to become best in class at the specific pain point that they solve. They also typically start selling to smaller businesses that feel that pain point acutely, and build/iterate with those customers. The barriers to entry are typically lower so there’s more competition between point solutions. IMO we’d see more innovation in the space if point solutions could actually flourish.

In other industries, the ability to integrate with other point solutions is actually a marketing point in order to be attractive to customers. Integrating with point solutions also helps to serve a wider group of customers that might want to bring their solutions with them. I run a small business and I use a hundred point solutions. They’re all quite good at what they do separately, and many of them make it easy to integrate with the other point solutions because users (who are also the buyers) demand that. 

Instead in healthcare, being called a point solution is basically the enterprise version of a racial slur. Companies buying from vendors expect fully featured solutions out of the gate or that they have a “good enough” solution for all of the other problems they’re facing. Or because they’re so complicated to setup and keep track of, middle layers like consultants or implementation people gain a disproportionate amount of leverage.

As a result, the history of healthcare company successes is almost entirely companies winning via contracting and distribution prowess rather than necessarily delivering the best-in-class product. And personally I don’t think that’s good!

AI Therapists

A few interesting things about consistently using an LLM for low grade self-therapy. Based on a conversation I had with a friend who channels unresolved mental health issues into memes.

1. The LLM has all the other non-therapy queries you've asked it - which can help give it a bunch of context about you. If it has more queries, maybe it can more quickly escalate you to a higher level of care if you exhibit more disturbing behavior.
2. It has your past "sessions" stored, so you yourself can go back and review things easily.  There are no issues with continuity of care and you can revisit your own sessions easily
3. There's no embarrassment in telling LLM things and it’s infinitely patient with you. Opening up to a human can be harder because it feels like they’re judging you. Plus you can talk about much smaller subclinical issues that probably don’t require a full visit.
4. You can converse at all times of the day. I bet for a lot of people the “intrusive thoughts” happen at night or when you’re particularly alone (e.g. on a plane).  Or maybe you feel very frustrated after a given conflict and want to work through it right then.

I know this sounds like the onramp to an incel manifesto, and I recognize that bots have A LOT of shortcomings today. They can even be dangerous, as we saw with the Character.ai case where a teen became suicidal after developing an emotional connection with an AI avatar.

But whether or not mental health professionals like it, people are going to use large language models to self-therapize. It’s a fraction of the cost and available 24/7. So we should think through ways it can be a part of traditional care.

For example, maybe a therapist and bot work in tandem. Khanmigo is a teaching assistant that acts as a bridge between teachers and students. Students can get help from a bot with strict guardrails to help work through assignments. Teachers get reports on where students struggled, which lessons need more deep dives, and additional context on students based on their bot interactions.

Similarly, it would be interesting if a therapist had their own bots with exercises, guardrails, etc. that patients could work with. Behavioral health professionals could expand patient panels and the sessions can dive into more specifics based on conversations with the bot. SlingshotAI seems to be doing something in this area by building a psychology specific foundation model.

This is going to get even weirder when people are wearing hardware recording them at all times. There’s a real possibility that people turn to AI companions to help them navigate scenarios or understand what happened retrospectively. 

Regardless of how you feel about this future, it’s around the corner so we should figure out how it can be shaped positively.

**Featured Job Posting**

As we work with more companies with our headhunting product, we’re thinking about more ways we can be valuable. So we’re going to feature one company/job we’re excited about working with.

Today that company is…OUT-OF-POCKET!!! New year’s resolution, we’re getting bigger this year.

Role: We’re looking for someone to help out with events and community management. We’ve been really focused on creating spaces for builders like hackathons, knowledgefest, the OOP slack, and more. And we need a hand!

If this is something you’d be interested in, here are more details about the job and how to apply.

Thinkboi out,

Nikhil Krishnan aka. “Men will literally justify therapy bots instead of going to therapy”

Twitter: ​@nikillinit​

IG: ​@outofpockethealth​

Other posts: ​outofpocket.health/posts​

--

{{sub-form}}

‎If you’re enjoying the newsletter, do me a solid and shoot this over to a friend or healthcare slack channel and tell them to sign up. The line between unemployment and founder of a startup is traction and whether your parents believe you have a job.