Why "AI Therapy" Is the Wrong Frame for Couples Tech

A category piece on the language we use for AI in mental and relational health. Why the word "therapy" is the wrong word for what AI can do, and why the right frame matters.

# Why "AI Therapy" Is the Wrong Frame, and "AI-Guided Wellness" Is the Right One

You have probably seen a lot of products in the last two years describe themselves as "AI therapists." The framing is appealing. It is also wrong. This post is the long version of why we, at PairLume, deliberately do not use that language, and why we think the distinction matters for users, for the field, and for the long-term legitimacy of any product in this space.

## What "therapy" actually means

Therapy is not a vibe. It is a regulated professional practice. In most jurisdictions, the word is reserved (with varying degrees of strictness) for services delivered by a licensed clinician with formal training in a recognised modality, supervised hours, ongoing continuing education, ethical obligations enforceable by a licensing board, and legal accountability.

When you walk into a therapist's office, you are buying not just the conversation but everything attached: the license, the malpractice insurance, the supervision, the diagnostic framework, the legal duty of care.

An AI cannot have any of those things, by definition. It does not have a license. It cannot be supervised in the legal sense. It does not carry malpractice insurance. The word "therapy," used about an AI, is at best loose and at worst a category error with regulatory consequences.

## Why the loose use of the word matters

Three reasons.

**1. It misleads users in a high-stakes area.** A user who believes they are receiving "therapy" from an app may delay or skip seeking actual professional care for issues that genuinely need it. Depression, suicidality, trauma, abuse: these need licensed humans. Any product that sits in a frame that blurs that line carries some responsibility for the delay.

**2. It invites regulators to act, broadly.** When products in a category over-claim, regulators eventually arrive. They tend to write rules that are conservative across the whole category, including the products that did not over-claim. We would rather operate in a category that polices its own language than one that gets policed externally.

**3. It cheapens the actual practice.** Real therapy is hard. Therapists train for years. The work they do is not interchangeable with any current AI. Calling an AI "your therapist" implicitly suggests it is. It is not.

## What an AI can actually do well

We are not arguing AI cannot be useful in this space. It can. Specifically, AI can be excellent at:

- Facilitating a conversation between two people in real time. - Providing a neutral, patient presence available outside business hours. - Naming patterns that the people inside the conversation cannot see from the inside. - Offering structured prompts, scripts, and frameworks. - Producing artefacts (summaries, decisions, open threads) that the people can refer back to. - Recognising when something is outside its scope and pointing the user to a human professional.

That list is genuinely valuable. It deserves a name. The name we use for it is "AI-guided wellness." Less catchy than "AI therapist." More accurate.

## Why "wellness" is the right word

"Wellness" is a deliberately broader, lower-stakes term. It describes work on general life, relationship, and emotional health that does not rise to the level of clinical care. Yoga is wellness. Meditation apps are wellness. Couples doing weekly check-ins is wellness. None of these things are therapy. All of them are useful.

Putting AI-guided couples sessions in the wellness category is honest. It positions the work for what it actually is: a daily-reps practice, not a clinical intervention.

## What changes if we all use the right frame

If the field collectively shifted away from "AI therapy" and toward "AI-guided wellness" (or similar), three things would get better.

**1. Users would have clearer expectations.** A user who knows they are in a wellness product looks for a clinician for the things wellness products do not handle.

**2. Builders would design more honestly.** Products in the wellness frame are forced to think about referral pathways, escalation, and integration with human care. That produces better products.

**3. The serious therapy applications of AI would have room to develop.** There are legitimate, careful uses of AI inside actual clinical practice (clinician-supervised, integrated into treatment plans, with appropriate guardrails). Those will get harder to develop responsibly if the consumer market poisons the well first.

## What this means for PairLume

We use specific language and we are deliberate about it.

- "AI-guided wellness session" instead of "therapy session." - "Wellness professional" instead of "therapist" in our directory pages. - "Live two-person AI session" as the core product name. - Explicit, in-product referral to licensed human professionals when the conversation crosses into clinical territory.

We will keep using this language even if the rest of the market does not. Because in our view, the long-term reputation of PairLume, and any other product trying to do something serious in this space, depends on getting this right.

## The bottom line

Words matter. "AI therapy" sounds like a feature. It is actually a category error. "AI-guided wellness" is less exciting and more accurate. We will take accurate.