Being Labs
Life Models
Being Labs is building Life Models: AI that follows one specific individual over time, learns what is normal for them, notices what is changing, and predicts what comes next.
A life is not a snapshot. It is a trajectory.
Foundation models understand moments. We are building models that understand individuals: a persistent, continuously updated state for each life, grounded in long-term multimodal observation and tested against real outcomes.
From moments to lives
Our first domain is pets: individuals we can observe continuously, with objective outcomes and low privacy risk. The same framework is designed to extend to education, virtual characters, and beyond.
- Individual state
Turn continuous video, audio, device events and owner notes into one updatable state per individual.
- Relative to their own baseline
Learn each individual's normal across days, months and full history, and detect deviations from it.
- Individual-conditioned prediction
Forecast state and risk at 1 / 7 / 30 / 180 days with calibrated confidence and a readable explanation.
- Heterogeneous experts
One central multimodal model reads latent tokens from knowledge-tracing, time-series, graph and research experts.
Being Labs Research
We publish what we are testing, how we will test it, and where the evidence stops. Every claim is marked as external evidence, a hypothesis, or a result.
Life Model
How we define the problem, design the model, and plan to prove or disprove it.
Explore →
Cat Recovery Lab
Can an individual's past recoveries predict the next one? Run it yourself.
Explore →
Evidence Map
What our past work already supports, and the gap we must close ourselves.
Explore →