While much of the public conversation is about whether AI can replace therapists, the more practical question for most providers is how to begin applying AI in a way that is safe, effective and aligned with their purpose. Our new white paper, AI Adoption in Large Mental Health Services, outlines a clear, risk-aware approach for doing exactly that. It introduces a framework for identifying appropriate AI use cases and building the long-term organisational capabilities needed to deliver impact at scale.
Introduction
Mental health services are under increasing pressure. Access challenges, workforce shortages and administrative burdens continue to intensify. In this environment, artificial intelligence is emerging as a critical area of focus for service leaders.
While much of the public conversation is about whether AI can replace therapists, the more practical question for most providers is how to begin applying AI in a way that is safe, effective and aligned with their purpose. Our new white paper, AI Adoption in Large Mental Health Services, outlines a clear, risk-aware approach for doing exactly that. It introduces a framework for identifying appropriate AI use cases and building the long-term organisational capabilities needed to deliver impact at scale.
Where AI can create value safely
The paper suggests that the best place to start is with low-risk, non-patient-facing use cases. Examples include note taking, referral processing and administrative tasks. These use cases are already delivering measurable value across Tacklit’s partner organisations, with human oversight built into the process.
The next level of maturity involves patient-facing AI that supports, rather than replaces, human-led care. These include risk monitoring tools that trigger escalation protocols based on clinical rules, and AI-assisted triage workflows. In each case, the AI acts as a co-pilot within clearly defined boundaries.
The paper also discusses more advanced categories, such as AI-led stepped care for mild presentations. These opportunities require well-defined escalation pathways and careful oversight. A fourth category, involving unsupervised AI therapy, is acknowledged but not recommended given the current level of risk and limited evidence base.
Laying the groundwork for AI success
Beyond specific use cases, the white paper highlights the importance of system readiness. Many service organisations still operate in fragmented technology environments that constrain AI effectiveness. To realise AI’s full potential, providers must build integrated systems, structured data, and intuitive digital interfaces for both staff and clients.
Governance and vendor partnerships are also critical. Providers need clear policies on AI decision-making and monitoring, along with trusted technology partners who can support not only the platform, but the implementation and change journey as well.
The full paper includes a proposed AI opportunity portfolio, a framework for assessing risk and reward, and examples of how AI is already being used safely in leading mental health organisations.
