April and I started Spring Health 10 years ago today. When we did, mental healthcare was almost entirely analog. You called a 1-800 number. You worked your way through a phone tree. You waited weeks for an appointment, then left work and drove across town to keep it. The most personal experience in someone’s life was delivered through the least personal infrastructure imaginable.
The last decade was about closing that gap. We replaced phone trees with a platform. We gave people direct visibility into who their provider is, what they specialize in, and exactly when they are available.
That was the analog-to-digital decade. The next one is AI-native. The mental healthcare it produces will look almost nothing like what came before.
Where care still breaks
Digital solved real problems. Faster access. Better matching. Transparent provider information. Outcomes you could actually measure. The experience, though, was still built around the transaction, not the person.
You can see it in the most ordinary version of trying to get help.
They find a therapist.
They build trust.
They make progress.
Then they change jobs, move, switch health plans, or need a different level of care, and the thread breaks.
The system asks them to start all over again at exactly the moment they have the least energy to do it.
That is the problem that has stayed with us from the start. Even when care worked, it rarely stayed intact across the normal changes in a person’s life. Even the best solutions of that era, ours included, optimized the transaction. They could create access to a moment of care. They did not solve for the lifetime.
What the digital decade got right
We founded Spring Health on a conviction: mental healthcare did not need to be trial and error. We proved a new clinical model, Precision Mental Healthcare, that made care far more personal, identifying the treatments and providers that worked best for each individual. Every match, every outcome, and every signal fed a model that learns what works, for whom, and under what circumstances. Ten years and millions of patients later, the results are independently validated and peer-reviewed:
92% of patients reliably improved or recovered from depression or anxiety, and they get better 56% faster than any other clinical model.And that patient’s healthcare costs $1,796 per year less as a result
Those numbers matter because they show that better mental healthcare isn’t a vague aspiration or a nice to have. It’s here today, it’s working, it is saving companies money, and it improves over time as the system learns.
.jpg)
What AI-native actually means
The machine learning we built into Spring Health from day one was the first version of this idea. What is possible today is a generational leap beyond it.
The availability and affordability of general-purpose LLMs have revealed more demand for mental and emotional support than ever before. Millions of people are already turning to these tools every day for support, reflection, and guidance. LLMs have exposed just how often the formal system was unavailable, unaffordable, or inaccessible when people needed help.
Legacy mental health systems are responding by bolting AI onto the old system, to speed up workflows, summarize notes, and look more modern. What they lack is a model of care that:
Remembers context.Learns over time.Helps support continue across providers, coverage changes, and life stages.
That requires a different foundation: the one we’re building over the next decade.
Building lifelong continuity
You wouldn’t sign up for a 401(k) if you knew you’d lose the balance and start over from zero every time you changed jobs. That is exactly what mental healthcare has asked people to do for decades. Every coverage change, every job change, every move resets the clock.
AI-native architecture changes that. It makes lifelong continuity possible in two layers: a relationship that stays with you, and a thread that carries your story forward.
The relationship: Alma
Alma’s coverage is what makes it possible to keep the therapist you trust, even when your employer changes plans or you change jobs. With more than 170 million people worldwide now able to access Spring Health, we have the foundation in place to protect the progress a person has built, across the changes that have always interrupted it.
You stay with your care team. You stay with the relationship that is doing the work. The trust you have built does not reset to zero.
The thread: Guide
Guide is Spring Health’s AI that supports people across every stage of their mental health journey, and the part of this that is genuinely new. It carries the context of your care across providers, sessions, and life stages. It provides a continuous thread that stays with you, learns from you, and helps every interaction with care build on every one that came before. Not as an alternative to therapy, but as the kind of memory that has never existed in healthcare before.
Guide is still early, but it is already proving its clinical impact in one of the most fragile periods of any care journey, the first few weeks, when most people fall off. In our first peer-reviewed study, members using Guide attended 5% more therapy sessions in their first seven weeks, completed their second session sooner, and improved continuity of care from 57% to 60%.
People do not just need help finding care. They need help staying engaged with it long enough for it to work.
None of this came at the expense of the human relationship. Members using Guide reported no decline in therapeutic alliance with their providers. AI-native infrastructure does the work humans structurally cannot do, while preserving space for the work only humans can do. For a provider, that means showing up to every session with the full picture, not just what the patient remembers to mention.
Together
The relationship holds. The thread carries your story forward. Together, they make something possible that has never been possible in mental healthcare before: care that follows you, learns from you, and gets stronger the longer your story is intact.
The bar AI has to clear
LLMs revealed the scale of the need. They also revealed their limits. They are immediate, but they are not built for longitudinal care, clinical context, or mental health safety. They have shown both the size of the opportunity and the risk of getting it wrong.
If AI is going to play a meaningful role in mental healthcare, it has to be held to a real standard. That is why Spring Health co-developed and open-sourced VERA-MH: the first open-source AI safety benchmark for mental health. The bar should be public, independently verifiable, and applied to everyone in the category, ourselves included.
The next 10 years
The first decade of Spring Health was about making mental healthcare digital, measured, and precise.
The next one will make it continuous, lifelong, and AI-native.
That is what we have been building toward for the last decade. It is what the next decade of mental healthcare will require. It is what the next decade will demand from all of us.