Skip to content
Book a Safety Review

Know what you’ll monitor before your next model goes live.

We make clinical AI safer and easier to monitor.

So you can catch problems early and act with confidence.

How safeAI works: inputs (data, models, clinical context, risk priorities, evidence, policies and thresholds, and who to monitor) pass through safeAI across language, voice, image and structured data, producing evaluation, monitoring, safety signals, subgroup insights, alerts, migration analysis and governance evidence for doctors, auditors and governance teams.

Inputs

  • Data
  • Models
  • Clinical context
  • Risk priorities
  • Evidence
  • Policies & thresholds
  • Who to monitor
  • Language
  • Voice
  • Image
  • Structured data

Outputs

  • Evaluation
  • Monitoring
  • Safety signals
  • Subgroup insights
  • Alerts
  • Migration analysis
  • Governance evidence
  • Doctors
  • Auditors
  • Governance teams

AI failures are already clinical failures.

Source: AI Incident Database (opens in a new tab)

From guesswork to continuous assurance.

Without safeAI

Higher risk. Less visibility.

  • AI gets tested once before deployment.
  • Quality gets judged by vibe checks.
  • Generic benchmarks define AI quality.
  • Safety failures can go unnoticed.
  • Evaluations aren’t tailored to workflows.
  • Model providers become trusted sources.
  • Subgroup performance remains largely invisible.
  • Production changes rarely trigger re-evaluation.

With safeAI

Continuous insight. Greater impact.

  • AI gets continuously evaluated in production.
  • Quality reflects real-world clinical performance.
  • Hospitals define their own evaluations.
  • Safety-critical failures get actively monitored.
  • Models are tested across specific subgroups.
  • Independent engines verify evaluation evidence.
  • Drift and degradation trigger investigation.
  • Every decision creates an audit trail.

From data to safer care

A continuous loop of evaluation, monitoring, and learning — across every model, modality, and clinical use case.

The safeAI workflow
Continual learning Safer healthcare
  1. 01Ingest
  2. 02Define
  3. 03Evaluate
  4. 04Verify
  5. 05Approve
  6. 06Deploy
  7. 07Monitor
  8. 08Detect
  9. 09Investigate
  10. 10Act
  11. 11Learn
  12. 12Migrate

Your clinical teams decide what matters.

And when something changes, they know whether it matters.

Before go-live

  1. 1

    AI proposes

    Monitor candidates, subgroups, failure classes and sampling plans.

  2. 2

    Verification checks

    Backtesting and recomputation before anything is adopted.

  3. 3

    Clinicians approve

    Approval sits with the people accountable for care.

After go-live

  1. 4

    Assess the severity

    Thresholds and risk tiers set the alert level and confidence.

  2. 5

    Route and investigate

    The right clinician, model owner or governance team gets the cases and evidence.

  3. 6

    Act and learn

    Review, hold, roll back or recalibrate. Outcomes refine future monitors.

Evaluation runs on private, institution-specific datasets and challenge sets built from your own evidence. Data never leaves your premises.

Six families of monitors.

First we evaluate and set a baseline. Then our agents continuously monitor it, as per the guidelines you set.

Illustration: each monitor family tracks a signal inside bounds set at baseline. The drift monitor moves out of bounds and is flagged for investigation.

  • Performance & calibration

    Within bounds

  • Output quality

    Within bounds

  • Safety & uncertainty

    Within bounds

  • Robustness

    Within bounds

  • Subgroup equity

    Within bounds

  • Drift

    Out of bounds

Baseline bounds Monitored signal Drift and degradation trigger investigation Illustrative

Clinical safety is a multidisciplinary problem.

Clinicians, AI safety researchers, statisticians, engineers, governance leaders and product builders working on one problem: making clinical AI oversight scalable.

Core team

  • Lara Grosso

    Lara Grosso

    Harvard DrPH, Founder, CEO

    DrPH at Harvard in AI & patient safety. 10+ years in global health (World Health Organization, CHAI) across 7 countries.

    LinkedIn: Lara Grosso (opens in a new tab)
  • Luke Payyapilli

    Luke Payyapilli

    Infrastructure Engineering Leader, Harvard-MIT

    Software engineering leader focused on scaling cloud infrastructure. Has led product-driven engineering at multiple successfully acquired companies.

    LinkedIn: Luke Payyapilli (opens in a new tab)
  • Francesca Mussa

    Francesca Mussa

    Harvard PhD, Chief AI Scientist

    PhD, AIM Harvard–BGB. Statistician building clinically meaningful medical AI, grounded in physiology & patient care.

    LinkedIn: Francesca Mussa (opens in a new tab)
  • Tirtha Pani

    Tirtha Pani

    Member of Technical Staff

    AI Safety scientist building machine learning systems to improve outcomes for people with chronic disease.

    LinkedIn: Tirtha Pani (opens in a new tab)
  • Isabelle Schulte

    Isabelle Schulte

    Harvard MPH, Head of Business Development

    Leads business development, fundraising, hospital partnerships, and GTM strategy. 8+ years in healthcare operations and transformation.

    LinkedIn: Isabelle Schulte (opens in a new tab)
  • Filippo Bargagna

    Filippo Bargagna

    AI Safety Scientist | Research Fellow @ UNIPI

    Supports the technical development of the safeAI platform and its alignment to the clinical workflow. Research Fellow on ML and Signal Processing at UNIPI.

    LinkedIn: Filippo Bargagna (opens in a new tab)
  • Sanjana Patnam

    Sanjana Patnam

    AI Safety Engineer Intern

    Researches clinical AI failure patterns and builds evaluation frameworks for safeAI. Graduate student at Northeastern University in Data Analytics Engineering.

    LinkedIn: Sanjana Patnam (opens in a new tab)
  • Tajvir Chahal

    Tajvir Chahal

    AI Safety Engineer Intern

    Building bias/equity audits for safeAI’s clinical AI platform. Background in data analysis; studying Machine Intelligence at Northeastern, with a B.S. in Computer Science from Rutgers.

    LinkedIn: Tajvir Chahal (opens in a new tab)
  • Jonathan Lee

    Jonathan Lee

    Business Development and Fundraising Intern

    Supporting the development of business operations and fundraising of safeAI. Undergraduate student at Northeastern University majoring in Finance.

Clinical partners

  • Mayo Clinic

    Clinical validation partner

    Ranked the #1 hospital in the world by Newsweek for eight consecutive years and a perennial U.S. News & World Report Honor Roll hospital.

  • Humanitas AI

    Applied clinical AI research

    The AI Center of Humanitas Research Hospital, a leading Italian academic hospital network spanning 12 hospitals.

Advisors

  • Brian Anderson, MD

    Brian Anderson, MD

    CEO & Co-Founder, Coalition for Health AI (CHAI)

    A physician and digital health pioneer helping shape national standards for trustworthy, safe and effective AI in healthcare. Previously Chief Digital Health Physician at MITRE and head of Informatics and Network Medicine at athenahealth. Harvard Medical School graduate.

  • Tom Pollard

    Tom Pollard

    Research Scientist, MIT · Technical Director, PhysioNet

    Technical Director of PhysioNet and a contributor to the landmark MIMIC clinical databases. Instructor in Biostatistics at Harvard T.H. Chan School of Public Health, with a PhD in computational modelling of patient physiology.

  • Cyrus Massoumi

    Cyrus Massoumi

    Founder & CEO, Dr. B · Founder & Former CEO, Zocdoc

    A healthcare entrepreneur and investor who founded Zocdoc and went on to build Dr. B. Founder and Managing Partner of humbition, and a Senior Advisor to Columbia University’s Mailman School of Public Health.

Awards and accelerators.

  • Winner · 2025

    Harvard Innovation Fund Competition

  • Winner cohort · 2025

    Harvard HealthLab Accelerators

  • Semifinalists · 2026

    President’s Innovation Challenge

See it on your own models and data.

Bring us a model, a dataset, or a production monitoring problem. We’ll map the evidence, risk, failure modes, and oversight workflow.

Book a Safety Review

AI compliance advisory.

Rules for clinical AI are changing on both sides of the Atlantic. We help hospital teams work out which requirements apply to each system, and what evidence to keep.

EUEuropean Union

  • EU AI Act duties for high-risk AI
  • Medical Device Regulation for AI software
  • GDPR for health data

USUnited States

  • FDA expectations for AI-enabled device software
  • HIPAA privacy and security
  • Federal and state rules on bias and transparency
Get in touch about compliance

Requirements depend on the system, its intended use and where it is deployed.