Why most wellness apps
don't actually help

There are now over 20,000 mental health apps available on the App Store alone. Downloads have tripled since 2020. And yet, population-level mental health has not improved. In the UK, anxiety and depression rates remain at record highs. Something is not working.

We spent several months reviewing the research on digital mental health interventions — what works, what doesn't, and why. What we found surprised us, and it directly shaped how we built SerenMind AI.

The problem with generic tools

Most wellness apps offer the same toolkit: mood tracking, guided meditation, journaling prompts, breathing exercises, and affirmations. These aren't useless — they can provide real short-term relief. But they share a fundamental problem: they are impersonal.

They don't ask what's actually driving your anxiety. They don't reflect back what your pattern of answers reveals about you. They treat all users identically, regardless of whether your stress comes from relationship conflict, financial pressure, a loss of purpose, or something more complex.

This is what the research consistently shows. A 2023 meta-analysis of digital mental health interventions found that personalisation was the single strongest predictor of user engagement and long-term benefit — more than the type of intervention, the frequency of use, or the technical sophistication of the app.

What actually creates change

The interventions with the strongest evidence share something: they start with a specific, honest assessment of where a person actually is. Not a generic mood score, but a nuanced picture of emotional state, stress sources, relationship quality, sense of purpose, and self-care patterns.

When people feel accurately seen — when the reflection they receive matches their internal experience — something important happens. They become more willing to engage with the harder questions. They trust the guidance more. They act on it.

This is why we built SerenMind AI the way we did. Five specific questions. Deeply personalised analysis. A report that names exactly what you're carrying, not what the average anxious person carries.

The role of AI

What large language models do genuinely well is synthesis and personalisation at scale. Given accurate, specific inputs, they can generate reflections that feel remarkably attuned to the individual — in a way that would require hours of skilled therapeutic conversation to produce manually.

We don't claim this is therapy. It isn't. But used carefully — with honest prompting, rigorous output validation, and appropriate guardrails — AI can produce something genuinely useful: a mirror that shows you clearly, and a map that helps you move.