image: ©sturti | iStock

Across NHS Talking Therapies services, the challenge is no longer identifying need. The challenge is meeting that need.

Demand for mental health support continues to rise. Referral volumes remain high, clinical resources are stretched, and services face sustained pressure to reduce waiting times while improving reliable recovery and improvement rates.

Against this backdrop, NHS services are increasingly exploring where AI can improve operational efficiency and patient outcomes, not as a replacement for clinicians, but as infrastructure that enables them to focus on care. Increasingly, that infrastructure is emerging across the care pathway, from referral and triage to treatment delivery and patient engagement.

Talking Therapies services receive around 140,000–150,000 referrals each month, equivalent to nearly 1.8 million referrals annually.

Yet the earliest stages of the pathway (referral intake, triage and initial assessment) remain heavily dependent on manual processes. Variability in referral quality, incomplete clinical information and administrative screening can delay treatment before it even begins. Operationally, this creates three persistent pressures:

• Longer waits between referral and assessment;
• High administrative burden on clinical staff; and
• Increased risk of inappropriate or ineligible referrals.

While national targets require 75% of patients to begin treatment within six weeks, waiting times vary considerably between services.

Delays matter. Earlier intervention improves outcomes in anxiety and depression, while longer waits are associated with higher dropout rates and poorer recovery trajectories. Improving triage efficiency has therefore become a critical opportunity to reduce wait times. The question becomes practical: how can friction at the first point of contact be reduced?

Several Talking Therapies services have begun addressing this challenge through AI-enabled digital triage.
By collecting structured clinical information directly from patients at referral, services can move from manual screening to data-led triage. Clinicians begin appointments with richer clinical insight already available, enabling faster and more consistent decision-making.

Research evaluating AI-supported clinical assessments has demonstrated measurable improvements. Studies examining Limbic’s intake and triage system demonstrate 93% diagnostic reliability across common conditions, including depression, PTSD and generalised anxiety disorder. The same research found 45% fewer changes in treatment allocation, suggesting more accurate initial triage decisions.

Digital intake pathways may also improve equity of access. Services implementing AI-enabled referrals have reported a 29% increase in referrals from ethnic minority patients and a 179% increase in non-binary patients accessing care.

For services managing rising demand and limited clinical capacity, improving the accuracy and efficiency of triage represents a significant opportunity.

Triage efficiency has also become a critical opportunity to reduce wait times. Evidence from Talking Therapies services and published research suggests that it can.

By automating structured information gathering at referral, AI-supported triage reduces the time clinicians spend collecting baseline assessment data. Clinicians begin appointments with a structured overview of symptoms, history and risk factors, allowing more time for formulation, treatment planning and patient engagement.

Mona Stylianou, Clinical Director at Everyturn Mental Health, described the impact: “Our clinicians are now better supported by the tools and systems they use. Every hour saved by Limbic Access means another hour can be spent providing treatment directly to the people we support.”

Improving information quality at the first point of contact supports better treatment allocation and outcomes. Even modest efficiency gains can translate into meaningful increases in clinical capacity.

The experience of Living Well Consortium, the NHS Talking Therapies provider for Birmingham and Solihull, illustrates how these approaches can work in practice. By 2023, the service was processing more than 11,000 referrals annually, a fourfold increase compared with 2019. To manage rising demand, Living Well Consortium introduced Limbic Access alongside existing phone and professional referral pathways. Within the first year, the system processed 42% of the service’s self-referrals, reducing pressure on administrative and clinical staff.

Accessibility also improved. Around 40% of referrals now occur outside working hours, allowing patients to access support at times previously unavailable. Data quality improved significantly: before implementation, only 35% of patients completed key demographic questions, compared with 98% after digital triage was introduced. Within six months, the service also reported a 14% reduction in patient dropouts and a 2.7x increase in patients identifying as LGBTQ+ accessing services, demonstrating how improvements in intake processes support both access and engagement.

Dropout rates, missed appointments, and disengagement remain barriers to reliable improvement and recovery. AI-powered tools are increasingly used to support patients outside scheduled appointments, helping them remain engaged with therapeutic activities.

Research evaluating Limbic Care has demonstrated:

• 23% reduction in treatment dropouts (JMIR);
• 15% fewer missed appointments (JMIR); and
• 40% fewer sessions required to deliver standard care pathways. (Limbic Research Team)

For clinicians, these tools provide visibility into patient progress between appointments, enabling more informed clinical conversations.

Naomi Holdsworth, Operations Service Manager at Bradford District and Craven Talking Therapies, explained:
“It gives people an additional way to stay connected and supported throughout their therapy. It’s part of our commitment to providing flexible, accessible support for everyone in our community.”

Charlotte Elliott, a PWP at the service, added: “It ultimately makes therapy more effective because we can overcome barriers and tailor techniques more closely to clients.”

By supporting patients outside scheduled appointments, these tools can improve outcomes while making more efficient use of limited clinical capacity.

Despite growing interest in AI across healthcare, concerns sometimes remain about whether technology risks replacing clinical judgement. In practice, AI systems function as structured support tools that help clinicians begin with better information while retaining full responsibility for assessment and treatment decisions. As one clinical lead put it:

“AI doesn’t replace the clinician. It simply ensures they start with better information.”

As AI adoption grows, rigorous evaluation remains essential. Research evaluating Limbic has been published in peer-reviewed journals, including Nature Medicine and JMIR, examining clinical outcomes and operational performance across large real-world patient populations.

For NHS services considering digital transformation, new technologies must demonstrate measurable improvements in both outcomes and service performance. Further research and published evaluations can be accessed here.

AI-enabled care is demonstrating meaningful improvements across the care pathway. For NHS leaders, the question is no longer whether AI will play a role in mental healthcare, but how it can be implemented safely and transparently to strengthen clinical care.

Used thoughtfully, these technologies offer an opportunity to strengthen the infrastructure supporting NHS Talking Therapies: improving access, supporting clinicians, and enabling services to deliver better outcomes
at scale.

For more information, contact: Kathleen Henrick, General Manager UK, [email protected]