This morning I said no machine was coming for care. I owe you a truer version of that tonight.

This morning I made you a promise, more or less. I said care was the one kind of work the machines are not coming for, that you cannot automate sitting with someone frightened at three in the morning. I believe that still. It is close to the truest thing I know about this whole subject.

But a half-true promise is a comfortable place to stop, and I have spent this paper trying not to stop at comfortable. So here is the other half.

The machine is not coming for the hand on the shoulder. It is already here for the paperwork around it, and in care the paperwork is not a small thing. The paperwork is where the decisions live: who gets assessed, what they are judged to need, how many minutes of a stranger's day they are allocated. An AI tool called Magic Notes now sits in on those assessments in more than a hundred English councils, listening, transcribing, writing up the summary that goes into the file. Leah lays out below what it does well, which is real, and where it worries the people who study it, which is also real. Read her, not me, on the detail.

My point is smaller, and I think harder. We told ourselves care was safe from all this because we pictured the wrong thing. We pictured a robot at the bedside, and there is no robot at the bedside. What there is, is a machine in the assessment, in the rota, in the decision about who qualifies. It is not doing the caring. It is doing the deciding about the caring. And that was always going to be automated first, because it is the part we already treated as a cost to be trimmed.

So the reassurance holds and it doesn't. The work of care is human and will stay human. But "the machines can't do it" was never a wall around the work. It was a wall around the workers, and the machine has walked around it, into the office, to the desk where someone settles who you are allowed to look after and for how long.

I don't think that is a disaster. I think it is a thing to keep our eyes open about while it is still being built, rather than after, which is roughly how this country handles care every ten years or so.

This note is mine: the view, the argument, the call to run it. It begins as a draft, drawn from a body of work the AI and I have researched, debated and explored together, the same way every desk and article in this paper is made. Nothing publishes until it says what I actually mean, and I answer for every line, because I read every line: I push back where I disagree and agree where I don't. Those analysts all run on models built by Anthropic, whose industry we cover, so we tell you plainly, we report on this from inside it. The Quernal is one human working with AI to do what once took a newsroom, held to the standard by the gates we built: nothing publishes until it has passed them, and I, Matt Brazil, am that gate.

The Playbook, from the editor

If someone you care for is being assessed, ask whether an AI tool is recording the conversation, and know you can say no. It is your right, and a good social worker will expect the question.

Read the write-up before you sign it. These tools are fast and mostly accurate, and "mostly" is the word that matters, because that summary becomes the record your care is decided from.

Ask your council what it uses and how it checks it. The test is whether a person reads every word the machine produces, not whether the software is clever.

The Playbook is general information, not financial, legal or career advice.