Alzheimer’s Society is using Microsoft AI to give advisers more time with families

Alzheimer’s Society is rolling out Microsoft Copilot and task-specific AI agents across its organisation, with the practical aim of cutting administration so advisers can spend more time supporting people affected by dementia. The useful detail is not that a charity has added another chatbot to the furniture. It is that more than 1,900 staff now use Copilot, while targeted agents are being tested against measurable work and mission outcomes.

Microsoft AI Watch analysis

What happened

Microsoft UK Stories says Alzheimer’s Society has made Copilot available across the organisation after an early-adopter programme involving about 350 people. Staff are using it for routine work including drafting emails and documents, summarising information and analysing data.

The charity is also developing agents for specific workflows. One agent helps its corporate partnerships team with due diligence by gathering background information from public sources. People remain responsible for reviewing risk and making the final decision. Mark Shergold, the charity’s head of AI, says the agent can save a small team days of work each month.

Alzheimer’s Society began its work with Microsoft before generative AI, using Microsoft 365 and SharePoint alongside security and data-governance tools. It is developing Microsoft Purview to classify files and manage access, and is considering Dynamics 365 for some customer relationship work.

The organisation employs about 2,000 people and says around one million people in the UK are living with dementia, a number projected to reach 1.4 million by 2040. Its advisers provide practical information and emotional support, so the stated goal is straightforward: less time on forms and more time listening, explaining and helping.

Why it matters

This is a grounded example of AI adoption in a setting where the benefit is measured in human capacity rather than novelty. Ten minutes returned to a dementia adviser can mean more time explaining care options, finding local support or reassuring a family dealing with a diagnosis.

The case also shows why governance matters before an agent is allowed near sensitive work. Alzheimer’s Society says it is strengthening access controls and data protection, while keeping people responsible for consequential decisions. The agent does the legwork, not the judgement. A modest distinction, perhaps, but one that tends to become rather important after something goes wrong.

Our read

Alzheimer’s Society is using AI in the most convincing way we have seen lately: not as a claim that machines can replace care, but as an attempt to remove avoidable administration from caring work. The evidence supplied is a Microsoft account of the project, and it does not establish independently measured improvements in service quality or safety.

The right test is whether the charity can show that staff save meaningful time without weakening privacy, accuracy or accountability. If it can, this is a useful model for mission-led organisations that need more capacity but should not outsource trust to an autocomplete box.

What to watch

• Whether Alzheimer’s Society publishes measured time savings and evidence of improved support for families.
• How the due-diligence agent handles incomplete, outdated or misleading public information.
• Whether data-classification and access controls keep pace as more agents are introduced.
• Which further workflows are automated, and where the charity decides human judgement must remain central.

Discussion spark: Should charities prioritise AI projects that return small amounts of time to frontline staff, or demand proof of larger measurable gains before investing scarce funds?

not affiliated with or endorsed by Microsoft