AI can draft a policy document in seconds, and the text will often read well. What it cannot do is maintain that policy when legislation or statutory guidance changes, map it to the nine Quality Standards your home is inspected against, or stand behind it when an inspector asks where a procedure came from and when it was last reviewed. Ofsted does not inspect documents. It inspects whether your home runs on a current, understood, evidenced set of procedures. That is the gap AI-generated policies leave open, and it is the provider, not the tool, who is accountable for it.

What AI-generated policies get right

It is worth being fair about this, because the appeal is real. A general-purpose AI tool will produce a structured, plausible policy on almost any topic in children’s residential care. The language is clean, the headings are sensible, and the cost is close to zero. For a provider staring at a policy set that has drifted out of date, that looks like an answer.

For a first rough draft of an internal working document, it can be a reasonable starting point. The problems begin when the draft becomes the policy.

Where AI-generated policies fail at inspection

They are not maintained. A policy is only correct on the day it is generated. Legislation, case law, statutory guidance and the inspection framework all move, and a generated document does not move with them. Within months, parts of it are quietly wrong, and nobody is watching for which parts.

They are not mapped to the framework. Children’s homes in England are inspected under the SCCIF against the nine Quality Standards set out in the Children’s Homes (England) Regulations 2015. A generated policy is not built against that structure, so when you need to evidence how your procedures meet a specific standard, the mapping work still has to be done by hand.

They have no provenance. When an inspector asks how a procedure reached its current version, “we generated it” is not an answer that builds confidence. There is no author, no review history, no professional judgement behind the wording, and no one accountable for its accuracy.

They are generic where your home is specific. AI produces the average of what it has read. Your statement of purpose, your children, your building, your staffing model and your local safeguarding arrangements are not average, and the points where a policy must be specific are exactly the points inspectors probe.

They can be confidently wrong. AI tools state incorrect regulatory detail with the same fluency as correct detail. In a sector where notification duties and timeframes carry regulatory consequences, a plausible error is worse than an obvious gap, because nobody catches it.

What inspectors actually ask about your policies

Inspection conversations about policies tend to come down to a few questions. Where did this procedure come from? When was it last reviewed, and against what? How do staff know about it, and can they describe what it means in practice? Does what happens in the home match what the document says?

Notice that none of those questions is about the quality of the prose. A beautifully written, out-of-date policy fails all four. This is why generating better documents does not solve the problem: the problem is maintenance, evidence and practice, not writing.

The alternative: maintained procedures

The reason providers subscribe to a maintained procedures service rather than writing in-house or generating with AI is that it moves the hardest job, watching every change and updating everything it touches, to a team whose full-time work it is. tri.x procedures for children’s services are written and maintained by qualified specialists, updated when the law or guidance changes, and mapped to the framework homes are inspected against, with the review history behind every procedure.

Whether that is right for your home depends on your size, your team and how you currently keep policies current. But the comparison worth making is not “AI draft versus paying for documents.” It is “who is watching for change, and what happens in my home when they miss something.”

Frequently asked questions

Is it against Ofsted’s rules to use AI to write policies?  No regulation bans it. But the registered provider remains fully accountable for the accuracy and currency of every policy, however it was produced. Ofsted’s interest is in whether your procedures are correct, current, understood by staff and reflected in practice.

Can I use AI to draft and then have someone check it?  You can, but be honest about the workload. Checking a generated policy against current legislation, statutory guidance and the nine Quality Standards requires the same expertise and the same monitoring as writing it, and the checking has to be repeated every time something changes. Most of the apparent saving disappears there.

How often should children’s home policies be reviewed?  Policies should be reviewed whenever relevant legislation, statutory guidance or the inspection framework changes, not just on a fixed annual cycle. A calendar review misses changes that land between review dates, which is why maintenance is a monitoring job, not a scheduling job.

What is the difference between tri.x and a recording system?  They do different jobs. A recording system holds your daily logs and care records. tri.x provides the policy and procedure layer, the written framework that recording operates within. Providers typically run both.