Surgical Safety Technologies is now Aimbient.Read more

Surgical AI Frontiers

The infrastructure behind surgical AI breakthroughs.

AI does not advance surgery on algorithms alone. It needs diverse, high-quality procedural data — and a trusted way to collect, de-identify, organize, and activate it across institutions and borders. Aimbient is that backbone: reliable global infrastructure that turns real surgical activity into the evidence, benchmarks, and model-ready data needed to push the field forward.

Explore the Initiatives

See the Trust Standard

THE BACKBONE

Before AI can improve surgery workflows and outcomes, the data has to be functional.

Simply training a model is not the most difficult part of operating within multicenter surgical AI. It is reliably moving thousands of sensitive recordings from contributors into a governed, usable data environment — all while maintaining privacy, consistency, and traceability along the way.

Aimbient handles the infrastructure between the procedure and the breakthrough, so collaborators can focus on the clinical question, the model, and the outcome.

01

Collect from anywhere

Import uploads from institutions, researchers, and clinicians into one dependable workflow — across geographies, devices, and varying technical environments.

02

Make privacy operational

Apply purpose-built de-identification to sensitive video and audio before they move into review, annotation, and model-development workflows.

03

Turn variation into usable evidence

Organize heterogeneous procedural data so expert reviewers and AI teams can work from consistent, research-ready material.

04

Keep humans in the loop

Support expert annotation, adjudication, peer review, and coaching — because clinically meaningful AI still depends on clinically meaningful judgment.

“Data diversity must be designed into the data pipeline from the beginning.”

Thomas G Weiser, MD MPH

Clinical Professor of Surgery, Section of Trauma, Critical Care, and Emergency General Surgery, Director of Global Engagement, Department of Surgery Stanford University School of Medicine

The algorithm may get the headline.
The infrastructure determines whether the work can happen at all.

The algorithm may get the headline. The infrastructure determines whether the work can happen at all.

SAGES CRITICAL VIEW OF SAFETY CHALLENGE

A worldwide stress test for safer surgical AI.

During laparoscopic gallbladder surgery, the Critical View of Safety is a recommended step used to confirm anatomy before key structures are divided. It is essential to preventing bile duct injury — yet it is not consistently achieved or assessed.

The SAGES Critical View of Safety Challenge, the first AI competition organized by a surgical society, asked international teams to build models that could assess this safety step from surgical video — and remain reliable across the clinical variability of real-world care.

Aimbient, then operating as Surgical Safety Technologies, enabled the video-collection and de-identification infrastructure behind the challenge. Aimbient processed almost 2,000 uploads from contributors in more than 50 countries. From that broader upload flow, organizers created a curated benchmark of 1,000 videos from 54 institutions across 24 countries, annotated by 20 surgical experts.

01 Upload flow

Contributor uploads processed by Aimbient

Aimbient de-identification

Almost 2,000

Uploads processed by Aimbient

50+

Countries represented across the upload flow

Illustrative. Points do not mark specific contributing sites.

02 Final Benchmark

Curated videos in the final benchmark

Institutions across 24 countries

Surgical experts annotating

03 Published outcomes

International teams tested against the benchmark

Relative gain in assessment performance achieved by the leading approaches

>0%

More than 80% reduction in calibration error achieved by the top-performing method

Relative improvement in robustness over the prior state of the art

The result is more than a leaderboard. It is a global benchmark — and a clearer path toward surgical AI that can handle uncertainty, clinical variation, and deployment beyond a single institution.

Read the Final Challenge Paper

WELLCOME LEAP SAVE

Building the systems to scale surgical expertise.

Over 300 million operations are performed every year, and yet an additional 143 million are needed to meet the basic health needs globally. Closing that gap will take more than training surgeons through traditional methods. New ways to assess skill, accelerate learning, and protect patients will be required in order to expand access.

Wellcome Leap's $50 million SAVE program — Surgery: Assess / Validate / Expand — brings together an international group of performers to pursue that shift. Its goals are deliberately ambitious:

Program goals

01 Expand the workforce

Demonstrate a pathway to train non-MD practitioners to deliver routine laparoscopic surgery with outcomes equivalent to MD surgeons in three years or less.

02 Accelerate expertise

Shorten MD surgeon training by a full year through tools that accelerate skill acquisition and objectively quantify competence.

03 Protect outcomes while access grows

Reduce postoperative complications and mortality by more than 50% through advanced sensing, monitoring, and pattern recognition.

Aimbient's Roles

As a selected SAVE performer, Aimbient, then Surgical Safety Technologies, provided platform infrastructure for researchers and academics working on new models for surgical training and objective performance assessment. Participating teams could upload de-identified surgical video, annotate key steps and events, and use AI-generated analysis to investigate risk and performance.

The partnership later expanded with the NIHR Global Health Research Unit on Global Surgery across seven countries — Benin, Ghana, Guatemala, India, Mexico, Nigeria, and Rwanda — using a shared video library to support training, peer review, coaching, and the development of safer surgical systems.

NIHR Global Surgery partnership

7 countries

BeninGhanaGuatemalaIndiaMexicoNigeriaRwanda

The learning loop

Upload01
De-identify02
Review03
Coach04
Improve05

Each case feeds the next

"This work would not have been possible without the transformative technology of Aimbient."

Thomas G Weiser, MD MPH

Clinical Professor of Surgery, Section of Trauma, Critical Care, and Emergency General Surgery, Director of Global Engagement, Department of Surgery Stanford University School of Medicine

When surgical expertise is scarce, every case should be able to teach the next one. Aimbient provides the infrastructure that makes that learning loop possible across borders.

Explore the Wellcome Leap SAVE Program

Read About the Expanded Global Partnership

De-identified by design

Privacy is not a gate at the end.
It is infrastructure from the start.

Privacy is not a gate at the end. It is infrastructure from the start.

Global research creates the diversity that better AI needs. It also multiplies the privacy, governance, and operational demands surrounding every piece of procedural media. De-identification has to be dependable enough for sensitive clinical environments — and scalable enough for thousands of files.

Aimbient applies purpose-built privacy controls across procedural video and audio so data can become useful without identity becoming the price of collaboration.

Visual identifiers

Faces are blurred in room video to reduce exposure of identifiable personnel and patients.

Off-field protection

Operative video is obscured when a camera is removed from the body, and light-camera feeds can be protected when they move off field.

Audio privacy

Detected PHI segments can be muted and voices pitch-shifted to reduce the risk of identification.

Governed activation

De-identified media can move into review, annotation, training, and research workflows within a controlled environment.

Visual identifiers

Faces are blurred in room video to reduce exposure of identifiable personnel and patients.

Off-field protection

Operative video is obscured when a camera is removed from the body, and light-camera feeds can be protected when they move off field.

Audio privacy

Detected PHI segments can be muted and voices pitch-shifted to reduce the risk of identification.

Governed activation

De-identified media can move into review, annotation, training, and research workflows within a controlled environment.

The SAGES Challenge paper explicitly acknowledges Surgical Safety Technologies for enabling the video-collection and de-identification infrastructure. This is privacy engineering proven in the work — not added to the story after the fact.

Explore the Trust Standard

Read the SAGES Challenge Paper

THE FRONTIER

The frontier needs a backbone.

The SAGES CVS Challenge tested whether surgical AI could assess a critical safety practice across global variation. Wellcome Leap SAVE is testing how technology can accelerate training, expand the surgical workforce, and protect outcomes as access grows.

Different questions. Different collaborators. The same infrastructure requirement: trusted procedural data that can move from the point of care to expert review, research, and AI without rebuilding the foundation each time.

SAGES CVS Challenge

Assess a critical safety practice across global variation

Wellcome Leap SAVE

Accelerate training, expand the workforce, protect outcomes

Shared infrastructure

Trusted procedural data

Point of care

Expert review

Research

AI

01 A benchmark that travels

Diverse data and rigorous evaluation help expose the gap between a model that works once and one that can work in the real world.

02 A learning system without borders

Shared, de-identified procedural media turns geographically dispersed expertise into an active resource for review, coaching, and discovery.

03 A foundation for what comes next

Once the data layer is dependable, research teams can move faster — from question to dataset, from dataset to model, and from model to meaningful clinical validation.

Aimbient does not just analyze what happens in a procedure.
It makes it possible for each procedure to advance the next.

Aimbient does not just analyze what happens in a procedure. It makes it possible for each procedure to advance the next.

Build the next surgical AI breakthrough on infrastructure already proven at scale.

For surgical societies, researchers, technology teams, and health systems working at the edge of what is possible, Aimbient provides the trusted data foundation to collect globally, protect identity, and turn real procedures into progress.

Explore the Clinical Evidence

Build with Aimbient

A world where every procedure advances the next.

Book a Briefing

One World Trade Center,
New York, NY 10007, USA

20 Queen St W, Toronto,
ON M5H 3R3, Canada

A world where every procedure advances the next.

Book a Briefing

One World Trade Center,
New York, NY 10007, USA

20 Queen St W, Toronto,
ON M5H 3R3, Canada

A world where every procedure advances the next.

Book a Briefing

One World Trade Center,
New York, NY 10007, USA

20 Queen St W, Toronto,
ON M5H 3R3, Canada