It listens to what you write, looks for what repeats underneath, and writes it back in words you can use. Here is each step, what it uses, and what it leaves alone.
Your childhood home, family, inherited beliefs, belonging, your online self, desire and relationships, work, escape habits, the inner critic, the future. Closed questions give structure; open ones give your own words.
The AI reads the whole intake at once and writes 12 short readings — how you love, work, want and defend yourself. Each names the pattern in plain words first, then the clinical term.
Each week you answer 8 questions. The AI summarises the week, rewrites all 12 readings, and suggests a next step: a conversation, a report, rest, or professional care.
In the Consulting Room Oneli answers in short replies, using your readings, your Catchups and your circle's answers. Each conversation closes with a short paragraph that stays in your Case File.
A written report pulls it together: priorities, a summary, the 12 readings with the evidence behind them, your relationships, how things are moving, and questions for a clinician.
A licensed psychologist designs Oneli for you, stays hands-on and receives the handoffs.
Psychodynamic work starts from one idea: what is hard to say directly shows up in what you repeat. The AI looks for four things.
The same scene in different years — the job that stalls at month nine, the partner who leaves the same way.
Topics that get shorter answers, a joke where a feeling would be, the question you skip.
What happens when someone gets close, and what you do next.
The words you reach for, the hedges, the sentence you start and change.
No DSM or ICD categories. A diagnosis is a clinician's work.
No IQ, no personality types, no scores of you. Reading Depth measures how much the AI has heard, not who you are.
It names what is, not what will happen.
Suggestions, never instructions. Your psychologist sets the plan; you decide what to do.
Oneli combines our own algorithms with large language models. Our algorithms structure your intake, follow your patterns from week to week and check every message for risk before Oneli answers. Language models write the readings and replies, in your language. Your psychologist decides what Oneli explores and what it leaves alone. We do not train a model on what you write.
If you are in danger, contact local emergency services now — 112 anywhere in the EU. TwentyThird is not an emergency service. How safety works
