Problem

Mantis addresses a core healthcare problem: individual-level data is fragmented, siloed, unstructured, and often too scarce to model rare conditions, unusual patients, or treatment responses reliably. Conventional statistical AI tends to predict populations rather than individuals, while periodic appointments can miss meaningful changes in cognition, behavior, and physiology between visits. The economic stakes are substantial: dementia imposed an estimated $781 billion burden in 2025, while depression cost U.S. employers an estimated $187.8 billion annually.

The clearest initial use case is continuous cognitive-health monitoring. Mantis’s Parva product is aimed at detecting changes in a person’s own baseline that may be missed by population norms or monthly clinical snapshots, particularly in conditions such as ADHD, long COVID, dementia, and perimenopause. Its broader platform also targets injury prediction, treatment-response monitoring, and other decisions where forecasting one person is more valuable than relying on an average patient.

Product / Service

Mantis builds living, predictive digital twins of individuals by combining clinical records, digital activity, biometric data, imaging, and other streams. Its system routes, validates, and synthesizes those inputs and uses models—including a physics-engine layer—to produce predictive representations of human physiology and behavior. The company positions the infrastructure as a consent-based connection: people control access to their data, while builders receive scoped, revocable predictions rather than raw data.

Parva is the most concrete current product: software installed on a laptop that passively reads signals such as keystroke latency, mouse movement, attention patterns, and response behavior, then compares each day with the user’s own historical baseline in a HIPAA-compliant database. It is distributed through cognitive-health and mental-health clinics, requires no wearable or invasive device, and flags risk rather than making a diagnosis while Mantis works toward FDA clearance. The intended benefit is a continuous record that can help patients and clinicians notice change earlier and make better-informed decisions.

Market

Mantis competes at the intersection of digital biomarkers, behavioral biometrics, healthcare digital twins, and synthetic biomedical data. The company’s positioning spans both software for continuous individual monitoring and infrastructure for medical research, clinical trials, procedure simulation, and predictive healthcare. Reported adjacent competitors in digital biomarkers include Mindstrong, Altoida, and Evidation Health; Unlearn AI is a larger neighboring competitor in clinical digital twins and synthetic trial data, although these companies generally focus on narrower signals or use cases.

Mantis is an early commercial-stage company rather than a clearly pre-revenue research project. It has raised $7.4 million in seed funding led by Decibel, with participation from Y Combinator and Liquid 2; TechCrunch reported that an NBA team was one of its main clients, and Forbes reported clinic distribution for Parva across ADHD, psychiatry, perimenopause, neurology, and related practices. The evidence does not disclose revenue, but it does show customer and channel traction, with FDA clearance, broader public release, and expansion into pharmaceutical research still part of the company’s commercialization path.