Now Available — Free to Download

Know yourself
without a single question.

MVAT Mirror maps your Big Five personality profile from the data you already create every day — your messages, notes, and writing. No surveys. No guesswork. No data leaving your device.

Free to start — no credit card required

Personality tests are broken.

They measure who you think you are in the moment you're being asked — not who you actually are.

🎭

Social Desirability Bias

You instinctively answer the way you want to be seen, not the way you actually behave. Tests measure self-presentation, not self.

📅

Retest Instability

Up to 50% of people receive a different type when retested weeks later. Real personalities do not change that fast. The test is just noisy.

Snapshot Problem

A 15-minute questionnaire captures one mood on one day. Your actual personality is a pattern across thousands of real interactions.

🔬

Low Predictive Power

Self-report scales correlate poorly with objective behavioral outcomes. Passive text analysis consistently outperforms them in peer-reviewed research.

Three steps to an honest self-portrait.

1

Connect Your Sources

Link the accounts you already use — X (Twitter), Facebook, and Instagram — via OAuth. Mirror keeps importing your history each time you open the app, until everything is in. No credentials stored.

2

On-Device Analysis

Mirror's NLP engine runs entirely on your device. It extracts linguistic markers — word choice, sentence structure, emotional valence — and maps them to the Big Five model without ever sending your text to a server.

3

Your Living Profile

Get a full personality profile complete with confidence intervals. Unlike a one-time test score, your profile updates as new data comes in, tracking genuine change over time.

Your data never leaves your device.

This is not a privacy policy promise. It is an architectural constraint. Raw text is processed into personality vectors on your phone. Those vectors — not your messages, not your words — are what Mirror stores. The raw data is physically unable to reach our servers.

  • All NLP inference runs locally — no text ever transmitted to servers
  • Only anonymized personality vectors are synced to encrypted cloud storage
  • Therapy and journal notes are stored exclusively in device-local encrypted storage
  • No third-party analytics SDK receives personal content
  • Full data export and permanent account deletion available at any time

By the Numbers

0 bytes

of your raw messages ever transmitted to our servers


100%

of NLP processing happens on-device before any sync occurs


7

personality frameworks available in Pro — all processed locally

What beta users are saying.

★★★★★
"I have taken the Big Five test three times and got a different result each time. Mirror's profile has been consistent across six weeks of use. That consistency tells me it is measuring something real."
Sarah M. — Product Designer, Beta Tester
★★★★★
"The privacy setup was the first thing I checked. Reading the architecture doc and confirming that NLP runs on-device was what made me actually trust it with my messages."
David K. — Security Engineer, Beta Tester
★★★★☆
"I was skeptical — personality from texts sounds like astrology. But watching my openness score correlate with weeks where I actually tried new things was genuinely surprising."
Priya R. — Researcher, Beta Tester

Testimonials are from beta program participants. Results vary.

Start free. Go deeper when you are ready.

Free

$0

forever

  • Big Five personality framework
  • Facebook and Instagram imports, free
  • 200 free import credits for X
  • On-device analysis
  • Basic profile view
  • Additional frameworks
  • Relationship mapping
  • Temporal tracking
Download Free

X imports use credits, priced at cost. X charges Mirror for every post its API lets us read, so we pass that through transparently instead of hiding it in a subscription: 1 credit per post, like, or bookmark of yours; 5 credits per mention (X charges five times more for those). Credit packs: 1,600 for $1.99, or 4,100 for $4.99 — enough for the deepest possible import. Credits never expire.

Honest answers to hard questions.

How accurate is on-device NLP compared to lab-grade analysis?
Computational linguistics research (Pennebaker et al.; Yarkoni, 2010) consistently shows that passive text analysis from natural communication outperforms self-report questionnaires for predicting real-world behavior. Mirror uses the same linguistic feature extraction methods validated in peer-reviewed research, adapted for on-device inference. All trait scores include confidence intervals — no result is presented as definitive below 0.7 confidence.
What data sources can I connect?
Today: X (Twitter), Facebook, and Instagram. Data is accessed via OAuth tokens — Mirror never stores your login credentials, and you can revoke access at any time. More sources (including Gmail) are in the pipeline.
Why do X imports use credits?
X charges Mirror real money for every post its API lets us read — roughly a tenth of a cent per post of yours, and five times that for mentions. Rather than hide that cost inside a subscription, Mirror passes it through at cost plus a small margin: 1 credit per post, like, or bookmark; 5 per mention. You start with 200 free credits, and packs cost $1.99 (1,600 credits) or $4.99 (4,100 credits). Facebook and Instagram imports are free because Meta's API does not bill per read. If you run out of credits mid-import, the import pauses and resumes automatically after you top up.
Does my data actually stay on my phone?
Yes — by architecture, not by policy. Raw text is processed into numerical feature vectors entirely on your device using on-device ML models. Those vectors — which contain no recoverable original text — are what gets synced to encrypted cloud storage for cross-device access. Your actual words never leave your phone. This is enforced at the code level, not through a contractual promise.
What is the Big Five model?
The Big Five (OCEAN) — Openness, Conscientiousness, Extraversion, Agreeableness, Neuroticism — is the personality model with the strongest scientific consensus. Unlike MBTI (which sorts people into 16 discrete types), the Big Five treats each dimension as a continuous spectrum and has well-documented predictive validity for career performance, relationship quality, and health behaviors.
Can I delete all my data?
Yes, at any time via the app settings. A full JSON export of your data is available on demand. Permanent account deletion removes your personality vectors from our servers within 30 days. On-device data is cleared immediately when you delete the app.
How is this different from other personality apps?
Most personality apps are questionnaire wrappers — they present the same survey instruments from mid-20th century research dressed up with modern UI. Mirror differs in three fundamental ways: passive data analysis instead of active questionnaires, on-device processing as an architectural privacy guarantee (not a policy), and a living profile that updates continuously rather than producing a one-time snapshot.

The most honest mirror
you will ever look into.

Free to download. No questionnaires. No data leaving your device.

Learn more at mvat.ai — or read the full launch post