Articles

What We Are Writing

Published thinking from our founders on what AI literacy actually asks of senior leaders, and where healthcare decision-making is falling behind the technology.

Featured

Published on INSEAD InTheKnow

Our founders write on AI literacy for senior decision-makers in healthcare.

INSEAD
INSEAD InTheKnowJune 2026 · Collaborative Post

Two GEMBA Alumnae, One Mission: Helping Healthcare Leaders speak Fluent AI

By Dianne Baunbaek and Joanna Reijgersberg-Siew, INSEAD GEMBA class of 2019

We both lost our fathers to prostate cancer. That loss shaped two very different careers, and seven years after graduating from INSEAD GEMBA together we have arrived at the same conclusion about where the constraint now sits.

The technology side of healthcare AI has stopped being the bottleneck. Adoption is happening fast and unevenly, with or without the conditions that would make it safe. What is missing is the literacy of the people who decide where, how, and on what terms AI enters the care system.

81%
of US physicians now use AI in practice
27%
have received no training at all
<1 in 10
APAC MedTech professionals hold both clinical and AI fluency
With commentary from Prof. Claudia Zeisberger and Prof. Sameer Hasija.Read on INSEAD InTheKnow →
From Cranberry Learn

Notes on Healthcare AI

Short, sourced pieces on the regulation, adoption data and buying decisions our clients are working through. Every claim is traced to a named primary source.

Regulation and compliance

Where the rules actually sit in Singapore and Europe, read from the primary documents.

Regulation · Singapore

Singapore's AIHGle 2.0: What Healthcare Leaders Actually Have To Do Now

MOH and HSA split accountability across developer, deployer and user. If you buy and run AI in your institution, you are the deployer, and that role carries duties you cannot pass to a vendor.

Read the article →Regulation · Singapore

What HSA's GL-04 Update Now Expects You to Document for AI Medical Software

Training dataset demographics, post-market drift monitoring, controls on continuous learning, and what a clinician should do when the model disagrees with them.

Read the article →Regulation · Singapore

Singapore's AI-SaMD Sandbox: What It Exempts, and Who Actually Qualifies

Public healthcare institutions only, Class A and B, non-critical conditions, consultant-level oversight, CEO endorsement, and patients must be told. Read the conditions before assuming it applies.

Read the article →Regulation · Explainer

Is Your AI a Medical Device? The Question Most Health Systems Answer Too Late

Intended purpose drives classification, not the underlying technology. Two systems built on the same model can land on opposite sides of the line.

Read the article →Regulation · Europe

The EU AI Act Deadline for Medical AI Moved. A Lot of Teams Are Still Planning Around the Old One

The Digital Omnibus pushed high-risk AI obligations to December 2027 and August 2028. If your compliance plan still says August 2026, it is built on a superseded timeline.

Read the article →

Evidence and adoption

What the survey and study data supports, and where the widely quoted numbers do not hold up.

Adoption · Data

Where Agentic AI Is Actually Landing in Healthcare, According to the People Buying It

61% of surveyed health system and health plan executives are already building agentic AI or have budget secured. Read the methodology before you quote the number.

Read the article →Evidence · Data

Ambient AI Scribes: What the Evidence Actually Shows, and What It Does Not

Burnout down 1.94 points, task load down 24.42, both p<.001. Also: 48 physicians, no control arm, and time savings that were perceived rather than measured.

Read the article →Leadership · Data

81% of Physicians Use AI. 27% Were Never Trained on It. That Gap Is a Leadership Problem

Physician AI use more than doubled since 2023. Training did not follow. 92% want more education and over a quarter have had none from any source.

Read the article →Practice · Adoption

Why Healthcare AI Pilots Stall, and What the Successful Ones Did Differently

A pilot designed to demonstrate feasibility will demonstrate feasibility. That is not the same as answering whether you should deploy.

Read the article →

Buying and implementing

The decisions that determine whether a deployment is worth anything two years later.

Practice · Procurement

Twelve Questions to Ask a Healthcare AI Vendor Before You Sign

The questions that separate a real answer from a confident one. Validation, population fit, drift monitoring, failure modes, liability and what happens when you want to leave.

Read the article →Practice · Procurement

Build or Buy: How Healthcare Organisations Should Actually Decide

Build and you take on the developer role and its documentation burden. Buy and you inherit a roadmap you do not control. Neither is cheaper in the way people expect.

Read the article →Practice · Strategy

How to Shortlist AI Use Cases When Everything Looks Promising

Four filters: is there a measurable baseline, does someone own the outcome, what is the cost of being wrong, and would anyone notice if it stopped.

Read the article →Practice · Measurement

Measure This Before You Deploy, or You Will Never Know If It Worked

You cannot reconstruct a baseline after go-live. Capture the process time, the error rate, the staff experience and the volume before anything changes.

Read the article →

Capability and governance

Who needs to understand this, to what standard, and why the usual answer is the wrong group.

Capability · Leadership

What AI Literacy Actually Means for a Healthcare Executive

Not a technical curriculum. The test is whether you can read a validation study, find its weakest claim, and ask the follow-up that exposes it.

Read the article →Capability · Governance

Train the Board, Not Just the Clinicians

Training budgets target the people already asking for it. The accountability sits with people who are not.

Read the article →

Explainers

Plain definitions of the terms that turn up in vendor conversations.

Explainer

What Is Agentic AI in Healthcare, and How Is It Different From a Chatbot

A chatbot answers. An agent plans, calls systems and takes steps. The governance question changes with it.

Read the article →Explainer

What Is Model Drift, and Why It Is the Risk Clinicians Notice Last

Performance degrades as populations, documentation practices and upstream systems change. Without deliberate monitoring, the first signal is usually an incident.

Read the article →

Bring This Thinking Into Your Organisation

The gap we write about is the one our programmes are built to close.

Start a Conversation →