Magic Notes (England)

Assisting social care workers through transcription and note-taking

The challenge

Social workers face growing administrative burdens, particularly with documentation and assessment write-ups, reducing the time available for direct client care. Traditional note-taking methods limit efficiency, as practitioners must complete detailed reports after visits to preserve personal engagement during sessions.

The solution

Kingston Council worked with Beam to create Magic Notes, an AI transcription tool for social care. It was developed within three months through close collaboration with social workers and occupational therapists, who tested early versions and provided feedback. The AI was trained on anonymised real-world data, with strong governance in place to mitigate potential risks. Furthermore, a Data Processing Agreement (DPA) was established with Beam, and a Data Protection Impact Assessment (DPIA) was completed.

The results

  • Practitioners reported time savings of 50-60% on administrative tasks such as completing case notes and assessments.
  • Time required for supervision write-ups dropped from 40 minutes to under 10 minutes.
  • Transcription accuracy exceeded expectations, reaching over 96%.
  • The pilot identified a potential 50% reduction in administrative time, enabling resources to be reallocated toward prevention services.

Learnings / insights

  • Different types of supervisions required different templates, highlighting the need for context-specific design.
  • The tool was found to be less effective for users with low verbal communication and scenarios with poor audio quality.

Details

Organisation name
Kingston Council
Government body
Local Government
User group
Wider public sector
Use case type
Reusable
Type of technology
Generative AI
Phase
Live
Impact
Better Customer experience / Improved efficiency / Time savings

Links

Get in touch

Email ai-knowledge-hub@dsit.gov.uk to:

  • find out how to use or collaborate on this tool
  • talk to us about your own tool

Content created: 18 May 2025 | Last updated: 12 June 2025