Simplifying Complexity: AI Easy Read Conversion

Every organisation that serves the public produces documents that some of the public cannot use. Consent forms, service letters, safety notices, policy summaries: all written for readers who process long sentences and formal vocabulary comfortably. For people with learning disabilities, and for many others, that assumption fails. The practical answer is Easy Read, and the practical question is how to convert complex documents to Easy Read without it taking weeks per document.

This post explains what Easy Read actually requires, why manual conversion has never scaled, and how an AI-assisted approach with human control changes the economics for councils, NHS teams, charities, schools and EDI leads.

What Easy Read is, and why it is not just plain English

Easy Read is a specific accessible format with its own conventions. A well-produced Easy Read document uses short sentences, one idea at a time. It replaces jargon with everyday words. It pairs each key point with a picture symbol that supports the meaning. And it uses a large, uncluttered layout with plenty of white space.

That last point matters: simplifying the words is only half the job. A paragraph of plain English in ten-point type is still inaccessible to the people Easy Read exists for. The format is the accommodation, not just the vocabulary.

In the UK, this is more than good practice. The Equality Act 2010 requires reasonable adjustments for disabled people, and the Public Sector Equality Duty asks councils and other public bodies to think about accessibility proactively, not just when someone complains. An organisation that publishes important information only in a format some service users cannot read is carrying a legal and reputational risk it may not have quantified.

Why manual conversion has never kept up

Most organisations that produce Easy Read today do it one of two ways. They commission a specialist agency, which produces excellent work slowly and at a cost that limits it to a handful of flagship documents a year. Or a motivated member of staff rewrites documents by hand, hunts for appropriate symbols, and wrestles with layout in a word processor, often over several days per document.

Picture a communications officer at a housing association. A change to the repairs reporting process needs to go out to all tenants, including around two hundred with learning disabilities. The standard letter takes an hour. The Easy Read version means rewriting every sentence, sourcing thirty picture symbols with usage rights, laying it all out, and getting it checked. Realistically it ships two weeks after the standard letter, if it ships at all. The people who most needed clear information got it last.

That is the pattern manual conversion produces everywhere: Easy Read as an occasional artefact rather than a routine output. The intent is genuine, the pipeline is the problem.

How AI helps you convert complex documents to Easy Read

askVERA is askelie’s Easy Read creation tool, and it treats conversion as a five-step process rather than a black box:

1. Upload the source. A PDF, a Word file, a web page URL, or pasted text. The complex document you already have is the starting point.
2. AI processing. The system restructures the content into Easy Read form: short sentences, plain words, one idea per line, matched with picture symbols from a built-in, auto-tagged Easy Read image library.
3. Review and edit. A person goes through the draft, adjusts wording, swaps symbols and corrects anything the AI has missed. The human stays in control of what the final document says.
4. Compliance score. Every document is scored against the EU Easy Read Guidelines, with 80% or above indicating a good standard. This turns “does this look right?” into an objective check.
5. Download and share. The finished document is ready to publish alongside the standard version, on the same day.

The compliance score deserves emphasis because it answers the question every EDI lead eventually faces: how do we know our Easy Read is actually Easy Read? A measurable score against a recognised guideline gives you something to point to in an audit, an accessibility statement or a board report. Multiple language support, currently in beta, extends the same process to communities who need simplified content in languages other than English.

Who needs this, and for what

The use cases follow the documents:

Notice what these have in common: the documents are recurring, time-sensitive and consequential. A consent form that a patient cannot understand is not merely unhelpful, it undermines the consent itself. These are exactly the documents where a two-week specialist turnaround is hardest to justify.

Keeping humans in control

A fair concern about any AI writing tool is accuracy: what if the simplified version changes the meaning? This is why the review step is not optional polish but the heart of the workflow. The AI produces the draft in minutes; a person who understands the subject confirms that “you may lose your tenancy” has not softened into something misleading, and that the chosen symbols support rather than confuse the message.

This mirrors the wider askelie position: AI does the heavy lifting, humans keep authority over the output. For public bodies especially, being able to show that every published Easy Read document passed through human review and achieved a documented compliance score is what makes AI-assisted accessibility defensible rather than risky.

What it costs to start

Pricing is deliberately low-friction. Personal plans start from £10 per document, with a five-document pack at £50 and volume rates falling to £7.50 per document. Teams that need to convert complex documents to Easy Read routinely, such as a council communications team or an NHS trust, can move to annual Teams plans from £2,200 to £5,000 per year. There is a free trial with no credit card required, so the sensible first step is simply to run one of your own documents through it.

A practical pilot: take the three documents your service users most often ask staff to explain. Convert them, review them with someone who knows the audience, check the compliance scores, and publish them next to the originals. You will learn more from that exercise than from any procurement questionnaire.

Accessibility as routine, not exception

The goal is not to produce one impressive Easy Read document. It is to reach the point where every significant document your organisation publishes has an accessible version as a matter of course, produced the same day, by the same team, to a measurable standard. AI-assisted conversion with human review is what makes that goal realistic, because it removes the bottleneck that made Easy Read a special project instead of a standard output.

The organisations your documents are hardest for are rarely the ones who complain loudest. They just stop engaging. Making information genuinely readable is one of the few accessibility investments where the benefit is immediate, visible and measurable.