Interdisciplinary AI Research

The science of AI that can perceive, reason, and act.

01 · The thesis

Four dimensions decide
whether a system can work.

01

Perception

Recognizing what is happening by ingesting real-time, multi-modal data inputs: video, audio, and sensor data.

02

Reasoning

Achieving correct, explainable conclusions that experts and auditors can both understand and validate.

03

Generation

Producing output that is accurate enough to be used directly in workflows with confidence.

04

Speed

Operating in real time and at the edge, inside the time that’s reasonable for the task.

02Research Domains

Expertise, engineered across six domains.

01
Computer Vision
Train computers to see, process, and understand images and video the way humans do.
02
World Models
Construct and operate simulations and digital twins that represent a physical environment.
03
Multimodal AI
Process and combine different types of data, like text, audio, and video, at the same time.
04
Neuro-symbolic Reasoning
Couple neural perception with structured knowledge and formal inference, so outputs are grounded, verifiable, and explainable.
05
Cognitive Science
Capture how experts perceive and decide, precisely enough to build into a model.
06
Robotics
Connect perception to physical action with the speed, safety, and confidence the real world demands.
03Proving grounds

We build the system each domain needs.

We work in high-reliability domains, where expertise is scarce and failures carry significant consequences. In each, we purpose-build AI solutions for the domain, embedded with the knowledge of its best experts.

Healthcare

Surgery & clinical care

Surgical robotics and operational workflows.

Industrial

Physical operations

Precision manufacturing and factory optimization.

Life sciences

Discovery & the lab

Scientific research in the dry and wet labs.

04The Engine

The machine behind the research.

A vertically integrated R&D engine: one organization that runs from clinical expertise and credentialed data to models, deployment, and IP.

50+Specialized experts, clinical + AI
25+Medical doctors
12PhDs in AI / CV / ML
2IIT partnerships
1M+Surgical Videos + Records
70+Surgical & Medical KOLs
01

Clinically credentialed data, at scale

Our video data is labeled by anatomists and physicians under senior clinical direction. Our labeling capabilities are not crowdsourced and are extendable.

02

Full-modality research under one roof

One organization ships video, MRI, and CT models covering anatomy, pathology, tools, and activity. Many groups specialize in one and outsource the rest.

03

An owned chain, research to deployment

The same team that builds a model also deploys it and runs it in production, so results move from the lab to live use quickly.

05Research frontier

Navigation still begins by pinning a frame to the bone.

Surgical navigation needs a fixed reference. Today that means screwing a rigid array into the patient's skeleton before the procedure can start. Removing it is a perception problem before it is a product one.

Method 01

Registration from anatomy

The reference is taken from the structures already in the surgical field. The system recognizes the anatomy it is navigating against, so nothing has to be attached to the patient to hold a coordinate frame.

Method 02

Tracking without a fixed array

Position has to survive a camera that moves and anatomy that shifts. We work on holding registration continuously through both, which is what a pinned array exists to guarantee.

In active development and not cleared for any specific indication. This research is carried into practice by DSS.

From research to delivery

Computer vision becomes Veris. World models become Movendi.

Explore the portfolio →

Research that reaches people.

If you want your work deployed, not only published, we should talk.