From EHCP to ISP, without retyping a word.

Every SENCO has lived this moment. A pupil arrives with an EHCP from their previous school, or a support plan a specialist wrote eighteen months ago, and somewhere in those pages is everything you need to know: the barriers to learning, the provision that was already working. The only way to get that into your own system has been to read the whole thing and type it out again, by hand, need by need.

We built ProvisionIQ to remove exactly this kind of work, so we shipped a feature that does the retyping for you. Upload the document, and AI drafts the ISP.

What It Actually Pulls Out

Drop in a Word or PDF version of an EHCP or existing support plan, on a brand-new pupil or one who already has an ISP, and AI reads it for four things:

If it's a new pupil without an ISP yet, AI also suggests a starting tier based on what the document describes, Universal through Specialist, which the SENCO confirms or adjusts. If the pupil already has an ISP, imported items are checked against what's already recorded and flagged where they look like duplicates, so staff can leave known overlaps unchecked rather than reviewing the same need twice.

How It Works, Step by Step

1. 1
   Upload the file. A .docx or PDF, on an existing pupil's ISP or when starting a new one. Both a word-processed EHCP and a scanned support plan work.
2. 2
   The text is scrubbed before anything leaves the building. The pupil's name and other direct identifiers are stripped locally. Staff see the scrubbed text and can edit it further before it's used.
3. 3
   AI reads it in the background. A full EHCP runs to tens of thousands of words. Processing happens as a background job so a long document never blocks the page or times out. A real EHCP in testing took about a minute.
4. 4
   Everything comes back for review. Needs, outcomes and provisions land in the same accept/reject screen used across the platform's other AI-assisted flows. Likely duplicates against what's already on the ISP are flagged so staff aren't re-adding what's already there.

“Nothing an EHCP says about a child gets saved to their record just because AI read it. A human still decides what goes into the ISP — the AI just does the reading and the retyping.”

Liam Wheldon, CTO — ProvisionIQ

Built Around the Pupil Data, Not In Spite of It

An EHCP is one of the most sensitive documents a school holds, so the scrubbing came first, not last. Before any text from an uploaded document reaches the AI model, ProvisionIQ strips the pupil's name and other direct identifiers, such as date of birth, NHS number, email address and phone number. Dates that describe targets or review points are kept, because they're the thing the feature is trying to extract in the first place, but nothing that identifies the child by name leaves the platform.

Staff can see exactly what scrubbed text is about to be used, and edit it further before it's sent, in case the automatic scrub missed something the original document phrased unusually. And because a real EHCP can run to twenty or thirty pages, we built this as a background job rather than something that holds the page open and waiting: staff can navigate away and come back to a progress banner on the pupil or ISP page, with the result waiting for them either way.

This isn't a separate system bolted on for one feature. It reuses the same evidence upload, AI-service and PII-scrubbing infrastructure that already powers ProvisionIQ's other AI-assisted suggestions, for outcomes, provisions and one-page profiles. One pattern, applied consistently, rather than a different set of rules for every AI feature in the platform. The full detail of what data goes where is public on our Sub-Processor Register.

Why This Matters More Than It Sounds Like It Should

It's easy to file this under convenience. It isn't. Re-keying an EHCP by hand is exactly the kind of low-value, error-prone administrative work that the White Paper is explicitly trying to design out of the SENCO role, and it's exactly the moment where information about a child is most likely to go missing: at transition, between schools, when a plan changes hands and the person receiving it is starting from a blank page under time pressure.

With ISPs about to become a statutory requirement for every child with identified SEND who doesn't have an EHCP, most schools are also sitting on years of informal support plans, one-page profiles and provision maps that were never in a consistent format to begin with. That backlog is exactly what AI document import can clear, not just the pupils arriving from other schools, without a term of manual data entry standing between a school and being ready.

“The goal was never to have AI decide what a child needs. It's to stop a SENCO's afternoon disappearing into retyping something that was already written down properly once.”

Liam Wheldon, CTO — ProvisionIQ

A SENCO still makes every judgement call. They still decide what's accurate, what's changed, and what a document from another setting got wrong. What changes is where their time goes: less of it lost to transcription, more of it spent on the review that actually needs a professional's eye.