Our co-founder and chair, Isar Mazer, joined a private What The Health panel on the future of mental health, facilitated by founder and Director Emily Casey. What The Health is a health tech media and community that curates analysis, convenes expert conversations and offers member-only content and events. Because this panel recording is only available to paying members, we are sharing an open summary so non-members can follow the key points.

Isar joined three expert leaders who approach the field from different angles. Professor Ian Hickie is co-director of Health and Policy at the University of Sydney’s Brain and Mind Centre and a long-standing national voice on mental health reform and digital innovation. Angela Lim is the CEO and co-founder of Clearhead, a workplace wellbeing platform, and trained as a paediatric doctor and researcher before building in digital mental health. Steve Duke is the founder of The Hemingway Group and author of The Hemingway Report; drawing on experience at McKinsey and two unicorns, he advises mental health organisations on how to succeed clinically and commercially.

The discussion opened by naming a hard truth. For years there has been a hard trade-off between access and quality. When services increased access, for example by putting more content online, they often lost personalisation and intensity; when they focused on high-quality, high-touch care, they reached fewer people. Steve set the tone by arguing the target should be better results for more people, not just more throughput, and that new models must stand up clinically and commercially if they are to last. Angela brought the client perspective into focus: access is only meaningful if people can enter care in a timely way and are matched to the right clinician. Ian noted variability in clinical delivery and the need to learn from models that reliably lift outcomes. Isar connected these threads, pointing out that several efforts to expand access have not consistently delivered better outcomes because personalisation suffers when the model is not designed for it.

The panel then asked whether the trade-off can be broken. The consensus was cautiously optimistic. Isar argued that the new technology and AI toolset, used wisely, can scale services while lifting outcomes, provided the model builds in triage, matching, escalation and measurement from day one. Ian pointed to self-care and community contexts as near-term wins supported by technology, where people can get help earlier and more often with clear guardrails. Steve returned to the quality lens: the promise of technology only holds if teams define what “better” means in clinical terms and prove it with data. Angela agreed on the potential but cautioned that generic advice and poor fit can set people back; systems need to make good matches and escalate safely when risk rises.

This led the conversation beyond “AI good” or “AI bad” towards a more practical question: which use cases carry which risks, and how do we design accordingly. Isar framed the landscape as levels of risk across categories of use. Some sit in lower-risk spaces such as administrative support, structured intake and psychoeducation. Others are higher risk and require human oversight throughout. Ian supported the idea that some scenarios can be done at relatively low risk, not no risk, and that this distinction matters for public trust. Steve added that quality gains appear when teams are explicit about which outcome they are improving and how they will measure it. Angela brought it back to safety and experience: if matching and escalation fail, people can be harmed even when the technology is advanced.

Everyone agreed that technology alone will not unlock the opportunity; incentives and measurement must shift. Ian argued that funding should reward outcomes rather than activity so providers can innovate around what matters to people, not just count visits. Isar emphasised the same goal while noting a pragmatic path in the present: innovators can prove value where incentives already align, particularly with insurers focused on prevention and employers focused on wellbeing and productivity. Angela observed that employer routes can speed adoption by embedding support where people spend most of their time. Steve underlined durability: if a model does not make clinical and commercial sense, it will not scale and impact will not persist.

By the close, a shared perspective had emerged. Access and quality can rise together when services are designed for fit from the start, with clear triage, matching, escalation and outcomes measurement. AI should support clinicians and care teams rather than replace them, especially where risk or complexity is higher. Funding and procurement need to catch up, and in the meantime there is room to test, learn and scale in settings where purchasers already value measured outcomes.

Our thanks to Emily Casey for a thoughtful, well-paced discussion and to What The Health for convening a high-quality forum. The full recording is available to WTH members. If you would like to watch the session and access further events and posts, you can become a member of What The Health on Substack.

This summary is authored by the Tacklit team. We have tried to correctly characterise the views of each panelist, but there may be inaccuracies. We hope this overview provides a useful window into what was a very rich discussion.