ORDO

Personal behavioral intelligence for AI

The behavioral layer for the AI people already use.

ORDO turns everyday AI conversations into a living model of a person — and uses it to help at the right moment, through the AI they already use.

Start a conversation Download one-pager
105 days of real daily use in an earlier personal version
42–45% next-action prediction on open time-use diaries, versus 23% for a frequency baseline
0.918 auROC reproduced hour-ahead behavioral-risk forecast; 0.875 week and 0.889 day

Research metrics are results on open datasets, not outcomes from the pilot.

The opening

Personal life has moved into AI conversations.

People already discuss fears, relationships, habits, ambitions and daily decisions with AI. But those conversations remain fragmented, reactive and locked inside individual products.

The old mental-health app model asks people to open another destination and fill in another tracker. ORDO does not.

01

Distribution inside existing AI

No new daily destination. ORDO is designed for MCP-compatible clients.

02

Model independence

The person keeps their memory when they change AI. Switching from one AI to another no longer means starting from zero.

03

Infrastructure value

ORDO can give AI assistants durable personal context and proactive behavioral support.

The system

Memory that learns. Support that arrives before the moment.

Specialized agents coordinate recording, reasoning, difficult moments, retrospectives and planning around a deterministic core.

Persistent memory

Personal RAG retrieves relevant episodes, commitments, emotional states, actions and outcomes from a person’s own history.

Personal predictions

Personal models estimate next actions and risk windows from longitudinal daily data. Forecasts are recorded before outcomes, calibrated with Brier scores and checked against baselines.

Proactive support

ORDO can restore a commitment, suggest a small action before a predicted risk window, help during a difficult moment and close the loop at night.

Deterministic core

Code owns the facts

Memory, calculations, forecasts and verification.

Language models

Models handle the conversation

Specialized roles for dialogue, reasoning and response judging.

Rules limit interruptions. When there is nothing useful to say, ORDO stays quiet.

Research foundation

Personal models built on longitudinal daily data.

Why this can win

A behavioral layer inside the tools people already use.

Personal context that persists

One durable memory across compatible AI clients, rather than fragmented context locked inside individual products.

Data generic models do not have

Longitudinal links between intention, context, action, outcome and what helped.

Support that closes the loop

From a morning commitment to a difficult moment and an evening review, within the same behavioral model.

Subscription model

Bring-your-own-key and bring-your-own-subscription tiers, with organization-specific onboarding later.

Where we are

A working personal foundation. A multi-user core in development.

  • Earlier personal version used for 105 days.
  • New multi-user core in active development.
  • Prediction research evaluated on open datasets.
  • Preparing a consented pilot with 5–10 people.
Business model Subscriptions with bring-your-own-key and bring-your-own-subscription tiers.

What we are seeking

Grant support + early-stage investment.

  • Grant: model evaluation and compute for the pilot.
  • Investment: complete the multi-user product, run live research and grow the longitudinal behavioral dataset.

Founder

Alina Akbasheva — psychiatrist, PhD in Social Medicine.

Recently completed her doctorate in Harbin, China. Building at the intersection of psychiatry, behavioral science, personal AI and longitudinal data.

nr.cor.aa@gmail.com WeChat: W15134626495