“AI personality analysis” covers very different products: some read your private chats, some read a paragraph you paste, some read a stranger’s LinkedIn profile for a sales team. This guide compares the main ones on what they read, which frameworks they score, what they cost and how they handle your data — including where they beat TypeMyVibe.
The short answer
The best AI personality analysis tool depends on whose personality you want read, and from what. To be typed from your own chat history across several systems, use TypeMyVibe. To read a relationship from a shared chat, MosaicChats is built for that and its full report is free. To get an instant MBTI and Big Five guess from any pasted text, the JustBuildThings Personality Detector is free and says it stores nothing. To profile colleagues or sales prospects at work, Crystal and Humantic AI are the tools made for it. And if you only want a casual read, pasting a few messages into ChatGPT or Claude costs nothing — with real limits, covered below.
We make TypeMyVibe, so weigh our view accordingly. Every competitor fact below comes from that tool’s own website, linked at the end, and was checked in October 2026. Prices change, so confirm them before you pay.
| Tool | Reads | Frameworks | Price | Best for |
|---|---|---|---|---|
| TypeMyVibe | Chat exports (WhatsApp, iMessage, Instagram, Telegram, LINE, KakaoTalk, ChatGPT, Claude, Gemini); public X or Reddit | MBTI, Enneagram + wing, Big Five (SLOAN), alignment | Free report; $8.99 / $29.99 upgrades (US) | Typing yourself from real messages |
| MosaicChats | Chat exports or screenshots | MBTI plus relationship metrics | Free full report; Pro $9.99/mo | Tracking a relationship over time |
| JustBuildThings | Pasted text, 500+ words | MBTI, Big Five | Free; paid plans $9–15/mo | A quick read of any writing |
| Crystal | Questionnaires, LinkedIn profiles | 16 types, DISC, Enneagram, Big Five | Free tier; Premium $49/mo | Workplace and sales |
| Humantic AI | LinkedIn URL or work email | DISC, Big Five | Free trial; paid plans | Profiling sales prospects |
| ChatGPT / Claude | Whatever you paste | Any you ask for | Free or your plan | A casual, conversational read |
TypeMyVibe: typing you from your own chats
TypeMyVibe reads a chat export — WhatsApp, iMessage, Instagram, Telegram, LINE, KakaoTalk, or your ChatGPT, Claude or Gemini history — or a public X or Reddit handle. You pick which person in the chat to analyse, and it returns an MBTI type, an Enneagram type with wing, a Big Five (SLOAN) code and a moral alignment, each with the lines from your messages it based the call on. The first report is free. Paid upgrades are $8.99 (blind spots, how others experience you, a shareable card) and $29.99 for the full report, which adds the cognitive-function breakdown (US prices; they are localised by country). A separate paid compatibility report reads a shared chat, or two people’s reports, for relationship strengths, friction and communication patterns.
Where it is weaker: for private chats it needs an export file, which is more effort than pasting a paragraph (public X and Reddit accounts only need the handle). The free report covers the verdicts and the evidence but not everything; MosaicChats gives away more for free. And on very long exports it reads an evenly spaced sample of the history rather than every message. Paid analyses keep the uploaded chat in private storage with the report until you ask for it to be deleted, and uploads are never used to train general models (see the privacy page).
MosaicChats: reading a relationship
MosaicChats takes exported chats from WhatsApp, iMessage, Instagram, Telegram or Messenger, or screenshots, and is built around relationships: a compatibility score, a sentiment timeline, response patterns, an activity heatmap, an MBTI read and an AI coach called Myrah. Its free plan includes the full report for every chat, with no signup needed to try it; Pro is $9.99 a month for more uploads and coach messages.
Pick it over us if you want an ongoing view of a relationship for free — sentiment over time, response patterns and a coach to ask — rather than a one-off personality and compatibility read. It does not offer Big Five, Enneagram or alignment, and its privacy policy says chats are kept while your account is active and that anonymised, aggregated data may be used to improve models.
JustBuildThings Personality Detector: instant, from any text
A free web tool: paste at least 500 words of anything you wrote — emails, essays, posts — and it returns an MBTI type with a confidence score for each letter pair, plus Big Five scores. The site says it keeps no database and does not store prompts, though the text is sent to third-party AI models to be read. Paid plans ($9–15 a month) unlock stronger models.
Pick it over us if you want a quick answer from writing that is not a chat, or you do not want to export anything. The trade-off is depth: one pasted sample, two frameworks, and no quoted evidence explaining the call.
Crystal and Humantic AI: reading other people at work
Crystal combines six questionnaires with predictions built from LinkedIn profiles and covers the widest set of frameworks here: 16 types, DISC, Enneagram with wing, Big Five, values and strengths. It has a free tier and Premium at $49 a month, and plugs into LinkedIn, Chrome and HubSpot. Humantic AI builds DISC and Big Five profiles of prospects from a LinkedIn URL or work email for sales teams.
Both are good at what they are for, which is understanding colleagues and buyers before a meeting. Neither reads your private chats, and both can profile people who have not taken a test themselves. For self-understanding they are the wrong tool.
Pasting your chats into ChatGPT or Claude
It works, it is free or already paid for, and you can ask follow-up questions. The weaknesses are structural rather than about model quality: long exports get cut off or only partly read; a raw export mixes every speaker, so you have to clean it up yourself; answers shift between runs and lean flattering; there is no scoring method or confidence level; and what happens to the text depends on your account’s history and training settings. A dedicated tool separates the speakers for you and samples evenly across the whole history instead of whatever fits in one prompt. Our AI MBTI test guide goes into why a quick chatbot read and a structured analysis differ.
Tools that look similar but don’t read text
Some “AI personality” tools read something else entirely. FirstVibe, for example, analyses a selfie for a first-impression “vibe score” rather than any writing. Most best-test lists are made of questionnaires — 16Personalities, Truity, the official MBTI and the free IPIP Big Five — where you rate statements about yourself. If you want a questionnaire, ours are free with no signup at /tests, including the Big Five (SLOAN) test and the Enneagram test with wings.
How to choose
You want to know your own type
Use a tool that reads a lot of your natural writing and shows its evidence. That is what TypeMyVibe is built for; a free questionnaire is the low-effort alternative.
You want to understand a relationship
MosaicChats reads the shared chat for sentiment, response patterns and compatibility, and its full report is free. TypeMyVibe's paid compatibility report reads it for strengths, friction and communication patterns.
You have writing but no chat export
Paste 500+ words into the JustBuildThings Personality Detector for an instant MBTI and Big Five guess.
You need to read colleagues or buyers
Crystal or Humantic AI, which build profiles from LinkedIn and work data. Not suited to self-discovery.
How accurate is any of this?
None of these tools has published an independent validity study for typing people from chat text, and that includes us. The research base is real but narrower than marketing suggests: language does correlate with Big Five traits in peer-reviewed work, with modest effect sizes. MBTI and Enneagram reads from text inherit those systems’ own weaker evidence. Treat any AI result as a well-argued hypothesis — which is why we show the quoted evidence, so you can disagree with it. Our study of 1,179 people typed from their messages shows where the systems agree with each other and where they don’t.