Hi friends,

We’re at an interesting point in the evolution of mental health tech.

Ten years ago, it was a small, emerging industry.

Today, there are multiple, well-established players. They are mature businesses, with significant provider networks, payer contracts and patient populations.

They’re fighting for market share and profitability, with many trying to get their business to a liquidation event in the coming years. To do this, they have been rapidly evolving their care model, building new technology and forging new partnerships.

2025 has been full of such announcements.

From the introduction of Spring’s Continuous Care model to Headspace’s AI companion and stratified care model, to Brightline’s opening of brick-and-mortar clinics.

There have been dozens more announcements like this, and honestly, it’s hard to keep track of who is doing what. But that’s what I’m here for!

Over the past few weeks, I’ve analysed each of these companies and the direction they are taking in 2025. I wanted to provide a clear view of the strategy being pursued by each business and to understand the major trends across the industry.

And so, in this edition of The Hemingway Report, that’s exactly what I’ll give you.

Let’s get into it.

📣 Apply to join our vetted community 📣

Applications for our vetted community are now open for THR Pro members. Over 600 people registered their interest in joining the community, and from today, we’re taking applications.

If you’re a THR Pro member, check your inbox to find a link to apply. If you are not a THR Pro member and want to join the community, you can become a member today for $29 / month. Click below to learn more.

I have to say, doing research for this article made me feel quietly optimistic about the future of mental health tech. Yes, a lot of these advancements in care models and services are not revolutionary. But as an ecosystem, we are finally, actually, delivering on them.

The promises of 2021 pitch decks have been turned into tangible products for real patients. And that’s great.

So, although some of the trends we discuss today may not seem new to you, what is new, is that these models and technology are now being implemented by some of the world’s largest mental health organisations. Organisations which are responsible for the care of millions of people. Yes, it’s fun to look at the shiny new innovations from young startups. But it’s perhaps more relevant to see how the major players are adapting their existing care models.

So let’s see what trends we’re seeing across the major players in Menta Health tech.

1. Service Expansion

If 2020 to 2024 was about expanding access, 2025 has been about expanding services.

Almost every mental health company has been adding new services to their offering. They all want a comprehensive care offering that they can sell to payers and clients.

In the month of April alone;

• Two Chairs added psychiatric services.

• Charlie Health launched a virtual substance use disorder (SUD) program, including group therapy, family therapy, and medication-assisted treatment (MAT).

• And Brightline launched their first in-person pediatric clinic in Brooklyn to supplement virtual care.

While everyone is expanding their service offering, not everyone is building it themselves. ThriveWorks partnered with NOCD to create an integrated care pathway for patients with OCD, combining general therapy with specialised ERP treatment.

In a more hands-off approach, Grow launched a Care Coordination program to allow therapists on their platform to refer high-acuity clients to vetted partners like Equip (eating disorders) and Charlie Health.

Everyone is expanding their range of services, and most major players now offer a fully comprehensive care model, at the very least having substantial therapy and psychiatry offerings.

Overall, I expect this trend to continue as comprehensive care becomes table stakes.

I’m quite interested to see how far this trend goes. What other services might these businesses want to add? Lifestance has expressed a desire to add psychiatric services like Spravato and TMS. Will any of the mental health tech companies follow suit? And what about if psychedelic treatments get approved?

A comprehensive care offering also allows businesses to implement stratified care models, another major trend for this market in 2025.

NB: Interestingly, one of the only companies that seems to have stayed in their lane is Talkiatry, remaining heavily focused on psychiatry.

2. Stratified, Personalised Care Models

“The right care at the right time”. This promise appears in 90% of all mental health pitch decks¹ .

But implementing this kind of care model has been a defining theme of 2025. Companies have been making conscious efforts to shift from one-size-fits-all solutions to stratified and personalised care. The idea is simple and compelling - give each individual the right level and type of support based on their needs, preferences, and risk profile. This should provide better clinical outcomes but also reduce costs.

So, how are businesses actually delivering on this?

In May, Headspace announced their own stratified care model - a vision for how they plan to provide better care for their clients. It uses clinical assessments and AI-generated insights to create a tailored care plan, which can consist of psychiatry, therapy, content journeys, meditations, and their AI companion Ebb - it will even help the user navigate their employee benefits. Interestingly, their data shows that 62% of Headspace’s enterprise members now engage with two or more modalities.

Also in May, Spring launched Journeys, personalised care pathways tied to life events. If an employee is, say, becoming a new parent or navigating a divorce, Spring’s Journeys provide a curated hub of relevant content, a recommendation of human providers at the right level (coach, therapist, or physician), and even specialty services for that individual. It’s a move away from generic programs to “deeply contextual care” that meets people where they are.

I think Journeys is pretty cool, actually. In particular, I can see how it could improve engagement, a problem for most employer-focused mental health/EAP solutions. A major life event can be a trigger for someone seeking support. If Spring can offer a journey to someone for the specific challenge they are going through, people are going to be more likely to engage with the service.

Within these stratified care models, what else is actually changing?

Breakthroughs in Triage

When I speak to clinicians, they tell me one of the hardest parts of any care model is triage. Headspace and others are betting that the combination of “clinically validated assessments with AI-driven insights” will unlock better triage.

To do triage well in mental health tech, you need a system that combines structured clinical assessment with contextual data (like life events, preferences, and prior care history). Running digital clinical assessments is not that hard. Gathering contextual data is more difficult. But a conversational AI companion like Ebb may be very good at this. If an AI-supported decision tool has access to this data, and good data on which it can learn, it could get very good at triage. AI-focused players like Limbic are heavily focused on solving these kinds of problems.

Unlocking better, more efficient triage is a huge unlock necessary for stratified, personalised care.

Dynamic Care Plans

Many of the new care models include continuous monitoring and adaptation of care plans. The idea here is that by including real-time progress tracking, goal setting and contextual data, as a member’s needs and situation evolve, the care plan can easily shift with them. Again, another good idea that has been around for a while. But AI is finally unlocking this capability.

For example, conversational agents (like Headspace’s Ebb) give organisations an easy way to keep both clinical and contextual data up to date. If a client is conversing with an AI agent, it’s much easier to gather contextual information that may otherwise have been ignored, and their plan can then be adequately personalised. Passive monitoring (e.g., with wearables) is not being used by anyone at scale but may hold potential. Sword just launched their mental health product Mind, which is betting big on this trend.

Hyper Personalisation

Everything we’ve discussed so far has essentially been focused on greater personalisation of care. But organisations aren’t stopping there. Hyperpersonalised provider matching continues to be a major area of focus, with Rula, Headspace, Lyra, Spring and more all introducing new capabilities in this area. I’ve always had mixed feelings about provider matching. On one hand, we know that the patient-provider relationship is very important. On the other hand, introducing strict matching criteria into a system that already has supply-demand imbalances is hard to implement for a business.

Many are also using Gen-AI to personalise content or, of course, to offer conversational agents - which are by nature, entirely personalised to the user. Talkspace even launched an AI-generated podcast based on clients’ therapy sessions.

3. Outcome and ROI infrastructure

Mental Health organisations have always thought about outcomes and ROI, but in 2025, this is becoming increasingly important. Payers are under cost pressure and questioning both the efficacy and return they are getting for behavioural health investments. In the US, many budgets are tightening. For mental health companies to maintain their contracts (and win new ones), they must be able to clearly demonstrate the efficacy and ROI of their service.

I think this is getting more important for clients, too. Over the last decade, there have been a lot of people who engaged with therapy, medication or other forms of mental health care, but haven’t experienced better outcomes. Some have been burned by bad experiences with mental health platforms. These people don’t just want access anymore. They want access to something they think is going to work.

To respond to both payer and client needs, the biggest mental health businesses have been investing in their data and measurement capabilities.

Spring built Compass, their own care delivery and EHR platform. Every Spring provider has to use Compass, and every provider-patient interaction is “captured in a single, unified system”.

Rula has embedded patient-reported outcome surveys (PHQ-9, GAD-7, etc.) into every visit via its own custom EHR.

Headspace’s stratified care model includes a significant focus on data capture.

From a branding standpoint, I think Lyra lead the way on communicating their commitment to quality and outcomes-based care. This year, they published their 20th peer-reviewed study, highlighting how their care model has “helped 9 out of 10 people get better and stay better”. They have also been investing heavily in their ability to track and report on outcomes data.

The ability to do this is not just important for today’s needs to convince payers and patients of ROI, but also for the longer-term strategy of any player in this space. If the industry ever gets genuine adoption of Value-Based Care, it will be the players with the most robust datasets and evidence bases (proving their outcomes) that will dominate. Finally, this data is needed to train the AI models that they all plan to use to support their care delivery in future.

4. AI for clinician support

Everyone’s looking for ways to implement AI in their business. However, client-facing AI applications are hard to bring to market. Clinician-facing tools have proved easier.

One example comes from Lyra. As part of Lyra Empower, they launched Engage, a “care delivery system that automates admin work, guides clinical decisions, and simplifies care coordination”. Talkspace are another example, launching Talkspace Insights, an AI-powered module that generates session briefs and summaries.

Grow, Headspace, Spring and many more have all announced new features that use AI to support their clinical teams. Again, these tools are quickly becoming table stakes for any major player.

5. A note on non-profit partnerships

Whilst not a care model or technical innovation per se, I’ve started to notice an interesting partnership trend. Mental Health tech companies are expanding their partnerships with non-profit players.

Non-profit providers make up a huge part of the behavioural health market in the US. A recent report from Behavioural Health Business outlined 50 of the largest behavioural health nonprofit organisations in the US. Every single one of them recorded more than $100M in revenue in 2023. Their combined revenue was over $17B!

They have deep roots and significant reach, but they sometimes have holes in their service offering or tech capabilities. In 2025, I’ve seen a few Mental Health tech businesses take advantage of this and forge clever partnerships with non-profit providers to expand the reach and breadth of their care.

Again, Lyra is a good example. They created a strategic partnership with Rogers Behavioral Health - a nonprofit provider - aimed at creating a full continuum of behavioural health care. The collaboration integrates outpatient therapy via Lyra with inpatient and intensive outpatient services from Rogers. Clever stuff.

So What?

The biggest players in mental health tech are now pursuing a remarkably similar strategy: build a comprehensive care platform that uses data and AI to triage users into the right level of care, deliver that care through a mix of human and technology supported services, and gather the data to prove it works and saves costs.

Regardless of whether a company started with therapy, coaching, psychiatry, or content, they are converging on a similar, comprehensive offering.

They are converging from a Go-To-Market perspective, too. Those who started direct-to-consumer now sell to employers. Those who started with employers now sell to insurers. And so on.

For these players, their new care models should allow them to outcompete legacy mental health providers. If the technology, data and breadth of services do everything they say they will, they should have superior outcomes (and an ability to show those outcomes).

Competition within the set of tech players will continue, however. If everyone has an AI-enabled, stratified, personalised, comprehensive care model, how do you differentiate yourself? This will be the major question on their minds. FWIW, I think distribution is the major strategic differentiator none of them have quite cracked!

For new entrants, these incumbents will be hard to compete with. A comprehensive service offering and advanced reporting capabilities are now table stakes. On the flip side, smaller businesses may become more attractive acquisition targets. As larger players look to round out their service offering, expand into new channels or differentiate themselves, they may decide to buy rather than build. Teladoc bought Uplift. Wysa bought April. Last year, Resilience Lab bought Options MD. The list is long, and I expect it to get bigger, especially as companies run out of runway and look for a soft landing.

Notes:

(1) Source: me

That’s all for this week. As always, reach out and let me know your own thoughts on this topic.

Keep fighting the good fight!

Steve

Founder of The Hemingway Group

P.S. Learn more about becoming a member of THR Pro