A clinical state summary that includes current biology, recent trajectory, and the next clinically bounded decision under review.
ANI.AI
A research lab building autonomous intelligence for human biology.
ANI.AI is a San Francisco research lab developing the computational substrate for longevity medicine. We integrate deep multiomics, longitudinal digital phenotyping, and physician-supervised intervention into a unified state model of the human body, designed to make individual response measurable rather than inferred from population averages.
What we are working on
Aging is the progressive loss of coordination between the systems that keep a person alive.
Muscle and metabolism, immunity and cognition, microbiome and host can each be measured in isolation, but the signal that matters for medicine is whether they continue to move together under intervention. When that coordination holds, function holds. When it breaks, single-system therapies tend to leave a patient improved on paper while function continues to decline.
Our work proceeds from a single hypothesis: that the coherence between biological systems is itself measurable, that it is the axis along which aging actually moves, and that a clinical practice capable of restoring coherence is achievable in this decade. The XPRIZE Healthspan target, twenty restored years of muscular, cognitive, and immune function, is the public expression of that hypothesis.
The platform
ANI is the research platform built to test it.
Each participant enrolled in an ANI study contributes a layered longitudinal record: deep molecular profiling at scheduled timepoints, continuous digital phenotyping between visits, validated functional and cognitive assessments, monitored intervention exposure, and a structured trace of physician decisions.
These signals are integrated into a learned latent representation of human biological state, against which individual trajectories can be inspected, compared, and reasoned about by both clinicians and machine agents.
Deep multiomics
Digital phenotype
Function + cognition
Intervention exposure
Physician decisions
A quantified picture of how a cohort moves through state space, including responder stratification, bottleneck identification, and propagation patterns between systems.
Machine-readable state objects that carry their own provenance, uncertainty, and operational constraints.
The platform is designed so that cohorts do not have to be large for the science to be deep. Sparse measurement is treated as a feature of real human biology, and inference is built around it.
Science
The coordinate system after clocks.
The work is organized around state, coherence, identity geometry, and the changing economics of measurement.
Beyond aging clocks.
Epigenetic and proteomic clocks compressed biology into a single number aligned with chronological time. They have been useful as benchmarks and limited as an interface for medicine. A clock cannot tell a physician whether a given system has been reached by a given intervention, whether response is propagating to adjacent systems, or whether an apparent improvement is the artifact of one domain pulling against the others. ANI is built around the question that comes after the clock: is this body becoming more capable of coordinated response, and where is that capacity being lost?
The Coherence Framework.
Coherence is our formal treatment of inter-system coordination. We quantify it through cross-system covariance across longitudinal multiomics, temporal alignment of response across functional and molecular layers, propagation depth from intervention to downstream systems, and resolution of counter-signals that point in different directions. The framework is in preparation for publication and serves as the analytic backbone of our trial work.
The Coherence Framework, forthcomingIdentity Mask.
The Identity Mask is a geometric representation of an individual's biological state. It is constructed from the latent space of the ANI model and exposes the structure of the manifold: where a person sits, in what direction they are moving, and what the surrounding axes mean, without exposing the raw underlying measurements. The Identity Mask is the object that clinical interfaces and AI agents query when they reason about a patient. The privacy-preserving geometry is the contract between the patient, the platform, and the systems that act on their data.
Read the paperThe shape of measurement is changing.
Slow molecular assays do not scale to lived biology. The future of longitudinal measurement is approximation: continuous digital and behavioral signal, anchored by deep molecular calibration at the rate of clinical decision-making. The platform is built around that asymmetry.
XPRIZE
Restored function is the endpoint. Response is the path.
XPRIZE Healthspan challenges teams to demonstrate restoration of muscular, cognitive, and immune function by at least ten years, with a full target of twenty. The FSHD Bonus Prize challenges teams to alter the course of facioscapulohumeral muscular dystrophy. ANI Biome was selected for both tracks in the May 2025 announcement, the only company in either field to enter both.
The prize measures restored function. Our work is to make the path to restored function measurable, individualized, and clinically governed.
XPRIZE HealthspanELITE
The first human substrate behind ANI.
ELITE is our healthspan intervention trial at Sheba Medical Center, conducted with the Sheba Longevity Center under the principal investigator Dr Evelyne Bischof. The protocol pairs physician-supervised intervention with deep longitudinal phenotyping: scheduled bloodwork, multiomics, microbiome sampling, validated functional and cognitive endpoints, and digital monitoring across the trial arc.
Each participant is studied as a trajectory of state across the intervention rather than as a row in a population average. ELITE is the first human substrate behind ANI. The trials that follow extend its design across additional cohorts and intervention classes.
Non-responders
Responders
Strong responders
Super responders
AgeBiotics
Defined inputs into the response loop.
AgeBiotics is the intervention layer of the lab. AB-01 is a liquid small-molecule matrix produced through controlled fermentation chemistry. AB-02 is a capsule cofactor formulation.
Both are studied as defined inputs into the ANI response loop, with effects characterized across microbiome ecology, metabolic flux, mitochondrial function, redox handling, and inflammatory signaling. Within ANI, every intervention is evaluated by the state change it produces in the individual who took it. That evaluation is the operational definition of N-of-1 medicine in our work.
Team
The people running the loop.
Bruno Balen
Co-Founder, Co-CEO
AI orchestration, response-state modeling, and the ANI platform.
After aging clocks, the next coordinate system is response. The question is no longer how old someone looks at the molecular level, but whether their biology is becoming more capable of coordinated motion under intervention.
Nika Pintar
Co-Founder, Co-CEO
Clinical deployment, protocol operations, regulatory architecture, bioinformatics.
Personalization begins the moment a protocol can remember exposure, tolerability, context, and motion. Anything earlier than that is averaging across people who do not move together.
Dr Evelyne Bischof
Principal Investigator, ELITE; Medical Director, Sheba Longevity Center
Longevity medicine becomes clinical when it is measurable, adaptive, and physician-governed. ANI gives the clinic a state model it can act on and a record it can be held to.
Prof Michael Snyder
Board, ANI Biome; Chair of Genetics, Stanford University
Longitudinal multiomics is where the individual trajectory becomes legible. Compressing that signal into something a physician can act on is where the field needs to go next.
Collaborate
Bring a cohort with heterogeneous response.
We work with research groups, clinicians, and translational teams studying aging, cognition, muscle, immunity, metabolism, microbiome, rehabilitation, and adaptive intervention.
Selected projects receive ANI platform credits, response-state modeling support, digital phenotyping infrastructure, and up to USD 5,000 per approved participant in credited multiomics services.
We are particularly interested in cohorts where response is visibly heterogeneous and the existing analysis cannot tell you why.
Waitlist