TL;DR
Angus Council has committed £240,000 to Magic Notes, a transcription system from technology firm Beam, to record and write up social work interviews. Councillors approved the move at a policy and resources committee in June, in private session, on the grounds that discussing it publicly could disclose the spend or terms. The award notice, now on the Public Contracts Scotland portal, shows no tender was advertised — the council cited an absence of competition on technical grounds.
What the council is buying
Magic Notes captures conversations, distinguishes between speakers automatically, produces summaries with action points, and translates sessions conducted in languages other than English. Beam says data is stored and processed within the UK. The tool carries UK government recognition and Angus joins a lengthening list of councils running it.
The stated rationale is time. A council spokesperson framed it as part of wider work on AI to streamline business processes, alongside existing supplier tools such as Microsoft Copilot. The pressure behind that is visible in the authority’s own numbers: figures released in June showed a £6m annual sick pay bill, with officials acknowledging absence was adding to the load on already stretched teams. A further £1m was approved last month for temporary staff on specific projects.
The procurement detail is the story
Two features of this award are worth separating from the technology. The first is the no-tender route: “absence of competition for technical reasons” is a legitimate justification under procurement rules, but it is a strong claim in a market where several vendors sell AI transcription into UK social care. The second is the private session. Neither is unusual on its own; together they mean a quarter of a million pounds of AI spend in a high-stakes statutory setting reached the public record only through the procurement portal.
Angus is not an outlier in direction. Lancashire County Council put Copilot in front of thousands of staff earlier this month. What differs here is the setting — social work assessments and interviews, where the record produced is evidence in decisions about vulnerable people, and where an automated summary that drops or reshapes a detail carries consequences a meeting note does not.
Looking forward
Councils adopting these tools are generating a body of practice well ahead of any inspection framework covering AI-assisted case records. The questions that follow — who checks a summary before it enters a file, how a disputed transcript is challenged, what happens when translation is wrong — are being answered locally, contract by contract, rather than nationally.