Why AI Therapy is Failing (and Wellbeing Companions Work Better)

There’s a fundamental tension in how mental health AI companies position themselves. Call yourself a therapist and you invite clinical liability, regulatory scrutiny, and the impossible claim of replacing human therapy. Call yourself a companion and you risk sounding like a toy.

Yet this positioning question is do-or-die. It determines what you build, how you’re regulated, and whether you actually help people.

Xuan Zhao learned this through experience. When building Flourish, she made an explicit choice: AI wellbeing companion, not AI therapist. And this choice is why Flourish succeeds while others fail.

The AI Therapy Trap

Most AI mental health companies start with the same assumption: “We’re building an AI therapist. We’ll use evidence-based therapy techniques — CBT, DBT, ACT.” These are legitimate techniques. They work. Duolingo-style apps use CBT effectively for anxiety management.

The problem isn’t the techniques. It’s the framing.

“There is a really interesting difference between being an AI therapist versus being an AI wellbeing companion,” Xuan explains. “We have been discussing, exploring this a lot with our clinical psychologists to what extent we wanted to actually administer some of these therapy techniques like CBT, right? Or to what extent we want it to be more doing interviewing, motivational interviewing and help people discover what truly motivates them.”

This choice cascades. If you’re an AI therapist, you’re treating people as broken and needing fixing. You’re positioning treatment, not support. You’re implying clinical equivalence to human therapy, which is both legally risky and factually false. And you’re inheriting the compliance burden of healthcare.

The Wellbeing Reframe

Flourish flips this. The positioning is: “You are fine. You can live a better life.”

This isn’t semantic wordplay. It’s a fundamentally different philosophy that changes the entire product.

“Oftentimes mental health apps take a deficit-based approach, right?” Xuan says. “They get inspired by therapy. But we know that in psychology, there is also positive psychology, which is about how to live a good life, even though you may not be clinically depressed or anxious.”

Positive psychology asks different questions than therapy psychology. Instead of “what’s wrong with you?”, it asks “how do you live better?” Instead of “fix the broken thing,” it asks “strengthen the foundation.”

This is why Flourish incorporates habit building alongside emotional regulation. This is why Sunny encourages you to reach out to friends instead of relying on the app. This is why the app is designed for people who want to flourish, not people who need to be fixed.

The business model shifts too. You can sell a wellbeing app to schools, employers, and healthy consumers. You can’t easily sell an AI therapist to those groups. Schools don’t want medical devices. Employers don’t want healthcare liability. Healthy people don’t want to be pathologized.

The Design Implications

When you design for wellbeing instead of therapy, the product is fundamentally different.

With therapy positioning, you’d build long, deep conversations. You’d try to resolve the root issue in a single session. You’d position yourself as a substitute for therapy.

With wellbeing positioning, you’d build for daily habits, connection to real people, and concrete actions. You’d position yourself as a foundation for living well. You’d explicitly NOT be therapy.

“We do not think people need AI friends to be completely honest,” Xuan says. “We think we need better human friendship with each other. We need to better connect with each other. So what Sunny does is Sunny oftentimes will make sure they will direct you to reach out to other people.”

This is the opposite of the engagement metric trap. An AI therapist wants you thinking you’ve gotten therapy from the app (engagement success!). An AI wellbeing companion wants you thinking the app helped you connect with real people (outcome success!).

The Regulatory Difference

Here’s where the positioning becomes a competitive advantage.

“We are providing Flourish as mental health promotion and early intervention. And we also have the social aspect of the Flourish app that helps people to connect better with each other,” Xuan explains. This positioning is legally distinct from therapy.

Mental health promotion isn’t regulated like therapy. Early intervention is a public health concept, not a clinical one. Connection is social, not medical. By repositioning the product, Flourish avoids the regulatory burden that crushed Woebot.

This doesn’t mean Flourish ignores safety. Actually, it does the opposite. “We have a tiered review system that we want to make sure we send some of the highest risk conversations to clinical psychologists to make sure that we review the conversation,” Xuan notes. They’re building safety infrastructure from day one, but not because they’re legally required to as a medical device. They’re doing it because the wellbeing philosophy demands it.

The Communication Challenge

The downside: it’s harder to explain what you do.

“Do you have anything else to add on like, you know, the messaging? And the philosophy?” Angelina asks. “What else do you think we’re different from like chat[GPT]?”

Xuan pauses. “It really depends on how you pitch this product to… even if it’s to healthcare, we are experimenting with a different sales.”

The positioning problem is real. When you search for “mental health app,” Flourish doesn’t show up. When you ask ChatGPT for mental health support, it recommends ChatGPT, not Flourish. The market defaults to therapy positioning because that’s familiar.

But Xuan has found the angle that works: “We talk about the value. AI or not, it doesn’t matter. Like at the end of the day, people want to feel better. They want to better habits. They want to have a safe place to organize their thoughts, to vent. And notice none of those involve AI, right? It’s the value that people care about.”

Lead with the value (feeling better, building habits, having a safe place). The AI is background. The positioning is wellbeing, not therapy.

The Long-Term Implication

This positioning also matters for durability. Therapy-positioned apps are vulnerable to being disrupted by better therapy. Wellbeing-positioned apps are harder to disrupt because wellbeing is foundational. You can always improve the therapy, but the foundation of living well is timeless.

“We want to set industry standard, standard of care of how to approach this responsibly,” Xuan says. “And there are already positive signs here.”

By staking the territory of wellbeing-first, Flourish is defining the category rather than competing in it. Others can claim to be AI therapy. Flourish is the standard-bearer for ethical AI wellbeing.

FAQ

What’s the difference between therapy and wellbeing in practice?

Therapy treats a condition. Wellbeing builds a foundation. Therapy asks “What’s broken?” Wellbeing asks “How do we live better?” An AI therapist tries to resolve your anxiety. An AI wellbeing companion helps you build habits and connections that reduce anxiety naturally. They’re philosophically different.

Why is positive psychology important for AI mental health apps?

Because most people using the app aren’t clinically depressed — they’re normal people wanting to live better lives. Positive psychology is for them. Therapy psychology is for people with diagnosed conditions. Designing for positive psychology is actually more inclusive (works for healthy people too) and less stigmatizing (you’re not broken).

Is Flourish therapy or not therapy?

Explicitly not therapy. Flourish is mental health promotion and early intervention. It’s designed to strengthen wellbeing and catch early signs of crisis, but it’s not therapeutic treatment. Xuan positions it this way intentionally to avoid liability and to target the broader wellbeing market, not the therapy market.

Can I use Flourish instead of therapy?

No. If you need therapy, see a therapist. Flourish is designed to complement therapy, not replace it. Many Flourish users are already in therapy — they use Flourish between sessions and to practice what they learned. But therapy with a human clinician is irreplaceable.

Why did Woebot fail with a therapy positioning?

Woebot positioned as AI therapy and bet on healthcare integration. Both positioning choices worked against them. Healthcare systems move slowly. Therapy regulation is heavy. Consumer adoption of “AI therapy” was limited because people prefer human therapists. The therapy positioning narrowed their addressable market while increasing their cost of sale.

Would Woebot have succeeded if it had pivoted to wellbeing positioning?

Possibly. Same product, different market. Instead of trying to sell to hospitals, they could have sold to schools and employers (Flourish’s path). Instead of claiming therapeutic equivalence, they could have positioned for wellbeing promotion. The product might not have changed, but the strategy would have been viable.

How do you market an AI wellbeing app when people want to talk to a therapist?

Don’t. Let therapists recommend it to patients. Position for schools and employers. Target healthy people wanting to live better. The people who want therapy will seek therapy. You’re building for the people who want wellbeing. Xuan says they “don’t even talk about AI” in marketing — they talk about feeling better and building habits.

Is this positioning just marketing, or does it actually change the product?

It changes everything. Therapy positioning pushes you toward long conversations, clinical precision, and handling severe cases. Wellbeing positioning pushes you toward daily habits, connection to real people, and early intervention. The architecture, feature set, and safety protocols all differ. It’s not rebranding — it’s a different product philosophy.

What about regulation? Is wellbeing positioning riskier or safer?

Safer. AI therapy is healthcare and brings FDA/regulatory risk. Wellbeing is public health and consumer-facing, with less regulatory burden. This is why Flourish can iterate faster and serve more markets. The tradeoff: you can’t claim therapeutic outcomes (but you prove them with RCTs instead).

Will Flourish eventually position as therapy to expand into healthcare?

Xuan doesn’t say, but she hints at it: “We are discussing with hospitals, cetera, but we are not providing Flourish as a therapy.” The long-term strategy is likely to stay as wellbeing positioning, but also to build safety and clinical evidence that hospitals eventually want to use. Different positioning for different customers — wellbeing for schools/consumers, clinical support for hospitals — but same underlying product.