AI therapy apps, compared honestly

We wrote this because the existing comparisons are unreliable. Checking the page-one results for “best AI therapy apps 2026” on 11 October 2026, several of them recommend apps that consumers cannot download — which suggests the lists are assembled from each other rather than from the apps.

What shut down, and when

Worth leading with, because it is the thing most comparisons get wrong.

What each one is actually for

The useful question is not which is best — these do different jobs, and the category label hides that.

Wysa — structured CBT techniques

The most clinically researched option still generally available, and the one closest to doing what the name “AI therapy” implies: it walks you through recognised CBT and DBT exercises rather than simply conversing. There is a paid tier and, in some markets, access to human coaching.

Choose it if you want technique — thought records, grounding exercises, structured reframing — and you want the option with the most published evidence behind it. Skip it if what you need is to talk something through at length; it is a guided tool more than an open conversation.

Ash, Earkick and the newer LLM apps

The post-ChatGPT generation. Much more fluent and open-ended than Wysa or Woebot, generally with less clinical validation behind them, and the category where the roster changes fastest. Earkick is usually recommended for mood tracking; Ash for open conversation.

Choose them if fluency matters more to you than structure. Be careful about the ones that lean hardest on the word “therapist” in their marketing — that framing is the least examined claim in this category, and the apps making it most aggressively tend to be the ones with the least behind it.

ChatGPT, Claude and general assistants

Worth naming, because a very large number of people are already doing this and no comparison page mentions it.

They are fluent, free or cheap, and genuinely useful for thinking out loud. What they are not built for: continuity between conversations, noticing a pattern across months, any safety handling designed for distress, or a privacy posture designed around this kind of material. Each new conversation starts from nothing — which is precisely the wrong property for anything you are working through over time.

Brihas — a thinking partner with memory

Ours, so read it with that in mind. Brihas is not a therapy app and does not run CBT protocols on you. It is built for the thing people actually bring: a decision that will not close, a pattern they half-see, a situation at work or at home they cannot say out loud anywhere else. It remembers across conversations, and it builds a reading of how you are navigating across six dimensions over time.

Choose it if continuity is the thing you are missing — if you are tired of re-explaining your life from scratch, or you want to see how something has moved over months. It also understands Hindi and Hinglish, which most of this category does not.

Do not choose it if you want guided clinical exercises — Wysa is better at that and is more researched. Or if you want a free app indefinitely: the first few messages and your first reading are free, and after that every plan is paid. Or if what you need is treatment, in which case none of the apps on this page is the answer.

Side by side

What none of them do

This is the most important section, and it is the one a comparison page has the least incentive to write.

• None of them can diagnose you, and none should try. If something needs a diagnosis, it needs a clinician.
• None of them is accountable the way a registered professional is. There is no licensing body, no duty of care, and no supervision.
• None of them handles real risk well. They are improving, but if you are in crisis the right move is a helpline or a clinician, today. In India, Tele-MANAS is 14416, free, 24/7, in multiple languages.
• None of them knows you the way a person does. A therapist notices what you avoid, what your face does, and what you did not say. That is most of the work, and it is the part that does not transfer.
• None of them is a reason to delay care you already know you need. The honest use of this category is for everything before and around that decision — not instead of it.

Where these tools genuinely help is the enormous space between “fine” and “needs treatment”: the ordinary, unglamorous business of being a working adult with decisions to make and nobody neutral to think with. That is a real need and it is badly served. It is simply not therapy, and the apps that blur the two do everyone a disservice — including themselves.