Empowering Neurodivergence Assessments

Project goal

Reducing clinical admin in children’s neurodivergence pathways in Wales

Project outcome

More engaged, high quality assessment appointments coupled with more efficient recording and outputs

Organisation name

Cwm Taf Morgannwg University Health Board

Department name

Neurodiversity

Organisation type

NHS Wales

Type of technology

Generative AI

Phase

Beta

What was the challenge?

As of December 2023, nearly 17,000 children and young people in Wales were waiting for an ADHD or autism spectrum disorder neurodevelopmental assessment, with ever increasing service demand hindered by pinch points such as specialist bottlenecks and difficulties accessing information. 

Who would benefit and how?

The Neurodiversity Pathway project aimed to address long waiting lists for autism and ADHD diagnoses in Wales caused by insufficient clinician time and exacerbated by repetitive administrative tasks. The project explored how AI tools could reduce administrative burden and pressure on clinicians and practitioners, making it easier for them to do their jobs while improving the experience for patients and their families, some of whom could have been waiting years for an assessment.  

What was the solution?

The team evaluated several AI scribe tools, prioritising ease of integration with clinician routines, vendor engagement and the ability to customise templates, before selecting Magic Notes by welfare-tech specialists Beam. Magic Notes is a generative AI tool that can; consolidate lengthy documents such as medical histories; listen to, transcribe and summarise clinical sessions; output information into a variety of templated reports that are compatible with NHS systems and processes.  

Key factors that helped clinicians embrace Magic Notes were the accuracy of the transcriptions and the wrap-around service provided by Beam, which made it easy to integrate the tool into existing routines and included unlimited template customisation. Ongoing collaboration with Beam’s dedicated team, responding to customisation requests, resolving technical queries and addressing security concerns, was essential in ensuring that the tool was fit for purpose and complied with governance standards. 

Who was involved?

The project team included clinicians, paediatricians, admin staff, information governance teams, digital directors and representatives from Beam, all of whom were engaged in a mixture of interviews, workshops and direct observation to gather insights, map the diagnosis journey and identify the pain points. The project’s agile approach to stakeholder engagement allowed for iterative testing and delivery to ensure the solution met user needs and complied with governance requirements. 

What was achieved?

Clinicians reported improved interactions with patients, as the tool allowed them to spend assessment appointments focusing on the patient and their family rather than administrative tasks. This led to richer and often shorter appointments and a reduction in time taken to write up notes and reports, in some cases saving up to 50% of a clinician’s time and reducing additional out-of-hours work. Long-term, this should have a positive impact on patient waiting times and clinician wellbeing. 

How was the Welsh language considered?

The team conducted tests with native Welsh speakers in Swansea University’s simulation suite in order to identify potential issues with transcription accuracy in Welsh. Due to an 11% error rate in Welsh, compared to 6% in English, we opted not to use Magic Notes for Welsh-language sessions, but it remains under consideration until either accuracy improves through further testing in the Welsh language or a better alternative is identified. 

What were the key governance steps taken?

Primary governance concerns, regarding ease of use and data security, were mitigated by an engaged and proactive relationship with the team at Beam. Early in the journey, Beam’s security team worked closely with the CTM information governance team to resolve data security concerns, and they ran multiple experiments to ensure the solution was relevant and that clinicians were comfortable using it while, financially, the cost of Magic Notes is well within typical procurement limits. 

What lessons have been learned?

Key to the pilot’s success was a recognition that the project couldn't solve everything. The initial period of stakeholder observation and collaborative process mapping was critical in identifying the true pain points which in turn created a focus on developing a relevant, effective solution. The project also highlighted the importance of early senior stakeholder engagement to remove any barriers to innovation and how a supportive vendor relationship can encourage practitioner adoption. 

Supporting information

Watch the project team’s ‘show and tell’ pilot review here 

Read the project summary here 

Read the ‘Making time for people, not paperwork’ blog here 

Read more about the AI Scribe testing phase in our blog  

Created By:  Matt Clark
Create Date: 06 Feb 2026
Last updated: 10 Feb 2026
Subject: AI » Generative AI