I've taken personality tests since I was nineteen. MBTI a handful of times, getting different results each time (INFJ and INTJ, depending on the week). Big Five through a university research platform during a study. The Enneagram during a phase when everyone I knew was using it to explain their behavior. The Attachment Project quiz. Truity's TypeFinder.
The results always felt approximately right in a way that didn't feel diagnostic. Like an astrology reading that happened to emphasize things that were already obvious — I tend toward introversion, I overthink, I care about ideas. None of the readings ever said anything that surprised me enough to make me change how I thought about myself.
Then I uploaded my ChatGPT export.
Why I wanted something different after years of MBTI and Big Five quizzes
The thing that eventually bothered me about self-report tests wasn't that they were inaccurate. It was that I couldn't trust my own answers.
I know the research on personality measurement. I know that conscientious people rate themselves as more conscientious than they are because conscientiousness is a desirable trait. I know that people with anxious attachment often rate themselves as more secure because they aspire to security. Every time I sat down with a questionnaire, I was aware — mid-answer — that I was probably shading my responses in some direction. Either toward how I wanted to be seen, or toward how I thought I actually was, which isn't the same thing.
When I read about Memrov using AI conversation history instead of questionnaires, the appeal was immediate and specific: my ChatGPT conversations were generated when I was trying to do things, not describe myself. I wasn't performing for an audience. The data couldn't be coached.
The export process: what it actually takes
Exporting your ChatGPT history is not complicated. In ChatGPT, go to Settings → Data controls → Export data. You'll get an email within a few minutes with a download link. The archive contains several files; the important one is conversations.json.
For an account I'd been using for about 18 months of fairly regular use, the file was 47MB. I have no particular sense of how representative that is — usage varies enormously.
The upload to Memrov took about two minutes. The processing took a few more minutes. Then I had a reading.
What the Memrov reading actually covers — all six frameworks
I wasn't expecting breadth. I expected a Big Five readout with maybe some interpretation. What I got was a reading that covered six separate frameworks and, more usefully, how they interacted.
Big Five: The results were recognizable. High Openness (I expected this; my conversation history has a genuinely bizarre range of topics), moderate-to-high Conscientiousness (higher than I would have rated myself, which gave me pause), moderate Extraversion, high Agreeableness, moderate Neuroticism.
HEXACO: The sixth dimension — Honesty-Humility — came in on the higher end. I hadn't thought much about this dimension before reading the description. The reading noted that this pattern, in combination with my Machiavellianism (Dark Triad, which we'll get to), produces a specific kind of strategic awareness that's self-constrained by ethical commitments. That framing felt accurate in a way that "you're a nice person" wouldn't have.
Attachment style: Mildly anxious-leaning. The reading was specific about the triggers: relational ambiguity in professional contexts (feedback that's neutral rather than explicit, unclear standing in group dynamics), rather than primarily romantic contexts. I've noticed this pattern in myself — I handle uncertainty in close personal relationships better than I handle it in professional evaluations where I genuinely don't know where I stand.
Schwartz values: Self-Direction and Universalism scored highest; Power and Conformity lowest. The reading noted a gap between my stated concern for universal welfare (Universalism) and my behavioral tendency to spend significantly more cognitive energy on self-directed projects and intellectual interests. This was the most uncomfortable finding. It's true.
Dark Triad: Subclinical Machiavellianism in the moderate range; narcissism and psychopathy both low. The Machiavellianism reading was careful — it characterized the pattern as strategic awareness rather than manipulativeness, and noted the combination with high Honesty-Humility as a meaningful moderator. This is the kind of nuance that a standalone Dark Triad test wouldn't have provided.
Motivation patterns: Approach-oriented, mastery-focused, primarily intrinsically motivated. This tracked.
The things it got right — and the one that surprised me
Most of the Big Five findings were recognizable without being particularly illuminating. I already knew I was high in Openness. I already suspected I was moderately conscientious.
The findings that actually changed something were:
The Conscientiousness score being higher than my self-estimate. I tend to tell a story about myself as disorganized and scattered. My conversation history, apparently, tells a different story — one of people who consistently breaks projects into structured phases and follows up on open questions. The gap between my self-concept (scattered) and my behavioral pattern (more methodical than I think) was a specific and useful correction.
The attachment read being about professional contexts specifically. I would have expected the anxious attachment signal to be more prominent in romantic relationship content. The reading noted that it appeared more strongly in professional-ambiguity conversations — uncertainty about how a project was landing, unclear feedback from collaborators, ambiguous signals about standing in a group. That's accurate in a way that made the finding harder to dismiss than a generic "you have some anxious attachment."
The values gap. Being shown explicitly that my stated Universalism coexisted with behavioral patterns showing significantly more investment in self-directed intellectual work than in the broader human welfare I claim to care about — that was uncomfortable. Not devastating. But specific enough to sit with.
Who should take Memrov — and who it's not for
Memrov is genuinely useful for a specific kind of person: someone who has already engaged with personality frameworks, is skeptical of self-report, and has enough AI conversation history to provide meaningful data. If you've used ChatGPT, Claude, or Gemini regularly for six months or more, you have enough data.
It's less immediately useful if you're brand new to personality frameworks — not because the reading won't be accurate, but because you won't have enough of a comparison point to know what to do with the nuances.
It's also not a replacement for therapy or serious self-examination. It's a highly specific kind of mirror — one that reflects behavioral patterns you weren't curating — but mirrors don't do the interpretation work for you.
For me, the most valuable thing wasn't any single finding. It was having a reading that I genuinely couldn't have gamed if I'd tried. That's an unusual feeling with a personality test — and it made the results land differently.
Upload your AI conversation history and get your free personality reading →