About
Just Human was built to solve a simple problem: in a world of AI-generated text, how do you show that you actually wrote something?
We record every keystroke, pause, and correction as you type, then turn it into a shareable replay. Anyone who clicks the link can watch your writing process unfold in real time — the thinking, the hesitation, the rewording. Things a machine doesn't do.
Why it matters
There's something happening to words. Every day, billions of them are generated instantly — polished, grammatically flawless, and utterly without effort. They arrive fully formed, with no crossed-out phrases, no pauses where someone searched for the right way to say something, no moments of doubt. They are perfect, and that perfection is exactly the problem.
Being human has never been about getting it right on the first try. It's about the backspace. The sentence you rewrote three times because it didn't quite capture what you meant. The long pause where you stared at the screen, thinking. The typo you caught — or didn't. These aren't flaws in your writing. They're proof that someone was actually there, wrestling with an idea, trying to make it honest. A machine can produce text, but it cannot hesitate. It cannot change its mind. It cannot care whether the words are true.
We built Just Human because we believe the process matters as much as the result. When you share a replay of your writing, you're not just sharing words — you're sharing evidence of thought. Every correction is a small act of integrity, every pause a moment of genuine reflection. In a world that's increasingly flooded with effortless content, the courage to show your rough edges might be the most human thing you can do.
By the numbers
How Analysis works
When you open a replay, the Analysis panel scores how human the writing process looks. It examines five independent signals from the keystroke data and combines them into a single 0–100 score. A higher score means the pattern is more consistent with real-time human typing.
- Rhythm Variability
- Humans don't type at a constant speed. This metric measures how much the gap between keystrokes varies — too uniform suggests a machine, while natural variation signals a real person.
- Error Corrections
- Backspaces and quick typo-fix sequences are a hallmark of human writing. The score looks at how often corrections happen and whether they follow the rapid delete-retype pattern people use instinctively.
- Paste Content
- Measures how much of the final text was pasted in rather than typed character by character. Small pastes (a URL, a name) are normal. When pasted content heavily outweighs actual typing, the score drops sharply — text that was mostly pasted in with little manual writing is a strong signal of machine-generated content.
- Think Pauses
- People stop to think — between sentences, before a tricky word, after re-reading what they wrote. This metric checks for natural thinking breaks and whether they fall near sentence boundaries.
- Timing Distribution
- Human keystroke timing tends to follow a log-normal distribution: most gaps are short, with a long tail of slower ones. This signal checks whether the overall statistical shape matches that expectation.
Scores fall into three tiers. Above 80 earns a “Written by a human” badge with a blue social preview. Between 50 and 80 the result is uncertain — the preview uses a neutral amber tone. Below 50, the analysis suggests the input may be automated, shown with a muted red preview. If a recording is too short to analyze — fewer than 30 events or under 5 seconds — it receives a neutral slate preview instead of being scored. In every case the full replay remains viewable.
No algorithm can definitively prove authorship. These signals are indicators, not guarantees — the replay itself is always the strongest evidence.
Built by
Development & research firm building tools at the intersection of finance, technology, and the open web.
dynamicstrategies.io →Get in touch
Have feedback, ideas, or just want to say hello? Reach out on X:
@dynamic_io