Insights · AI Adoption, Adult Social Care
Hey Geraldine made the national AI guidance. Here is what all our work has taught us.
22 November 2026 · Emma Ockelford
On 20 November the Government published its first national guidance on AI in children's and adult social care. Ministers chose two examples to announce it. One of them was our very first AI project - Hey Geraldine in Peterborough.
We supported its development, so that was a very good Friday! But the guidance matters to me for a bigger reason. It describes what we have learned everywhere else too: in our consultancy, in Plus Innovations™ and in the LGAi community we convene.
Five lessons stand out.
1. Start with a real problem
Geraldine Jinks has spent 35 years helping people in Peterborough live independently at home. Colleagues went to her constantly for advice. Working with Peterborough City Council, we supported the development of a no-code chatbot built on what she knows, and it was rolled out across the workforce. During its pilot period it answered more than 1,200 questions and saved around 15 minutes each time. Not only that but since the implementation of Hey Geraldine, the OT backlog has been cleared.
Nobody set out to "do AI". The team set out to solve a problem practitioners already had. The guidance asks councils to check that AI is of real benefit before they buy anything. That is the right first question.
2. Sometimes the answer is not AI
Hey Geraldine was one result of a wider piece of consultancy. Peterborough asked us to help increase the use of technology in adult social care. That became a digital roadmap, which then reached into public health, children's services, SEND and housing.
Most of what sits on a roadmap like that is not AI. It is better information, simpler processes and staff who feel confident. The guidance is honest that AI will not make up for information that is patchy or badly organised. We agree. AI where it fits, and no AI where it is not needed.
3. Judgement comes first
The practitioner guide has a short instruction I hope every director picks up: think first, then draft.
That is how CHC+™ works. It supports social workers with NHS Continuing Healthcare checklists. In Ealing, practitioners spend 80 to 90% less time on each checklist, and more than £1.3m of NHS funding was secured in their first year using CHC+, and for individuals with learning disabilities and physical disabilities for the first time. It only works because the practitioner's view comes first. They know the person, weigh the evidence and own every word. The guidance covers social care, not Continuing Healthcare, but the principle carries. Behind every checklist is an adult whose funded care depends on it.
I wrote in Community Care about how AI can support legal literacy in social care. This is the same point. Good tools make professional reasoning stronger and easier to see.
4. People should recognise themselves
SEND also sits outside this guidance, but one idea belongs in every SEND service. Add what AI cannot capture, and let people see what has been written so their voice is right.
EHCP+™ helps SEND teams produce review-ready drafts of Education, Health and Care Plans. The test we care about is whether a young person or parent can read the plan and recognise their priorities. A plan can hold all the right facts and still lose the point the family was making.
That is why every Plus Innovations tool comes with a clear standard and hands-on implementation support, not only a login. The guidance calls this treating AI as a change programme.
5. Nobody should work this out alone
The guidance asks for peer learning. LGAi, the independent community we co-founded and now run, now has more than 550 members across local public services, sharing what has worked and what has not. The most useful sessions are always the honest ones.
The question the headline leaves open
The announcement leads on time saved. I understand why. But time saved is a means.
For any council starting a pilot, I would ask practitioners and people drawing on support to name one thing they want the time back for. Then, when the pilot is reviewed, ask whether it happened. Time saved belongs in the business case. What people did with it belongs in the decision to scale.
The guidance is not statutory, and more will follow. What it does is give every council the same place to start. We are proud to have helped show what a good start looks like.
Because Outcomes Matter.