Upload a document. Get every gap, ranked.

The full journey, step by step

Six stages. Each one lists what you do, what happens out of sight, what you see, and why the step exists.

1. 01

   Upload

   A PDF or a Word file, dragged into a box. That is the whole setup.

   What you do

   Drag your RAMS into the upload zone, or click to browse. If there is something site-specific you want looked at closely — a revision that changed the scope, a section you already suspect is thin — you can say so in a note. Nothing else is required: no template, no field mapping, no account configuration.

   What happens in the background

   The file is accepted over TLS and stored in the UK. Text and structure are extracted — headings, tables, the hazard register grid — so that later steps are reading sections rather than one undifferentiated wall of text. Scanned or image-only PDFs are the one case that struggles here, which is why we say so on the limitations page rather than in a footnote.

   What you see

   The drop zone is replaced by a confirmation row showing the file was accepted and how many sections were detected. If the section count looks wrong, that is worth knowing before the review runs, not after.

   Why it matters

   Every review tool that asks for a template gets used on the documents that fit the template. RAMS documents in UK construction do not have a common format, so requiring one would mean reviewing only the easy half.

2. 02

   Analysis

   Each section judged on its own, then all of them judged against each other.

   What you do

   Nothing. This is the part you are not doing by hand. It takes about a minute and a half on a typical document, and you can leave the page.

   What happens in the background

   Two passes. The first reads each section against what a thorough reviewer would expect to find in it — is there a named muster point, does every hazard carry a residual risk score, is the control measure an action or an instruction to be careful. The second pass is the one manual review struggles with: cross-referencing sections against each other, so a hazard with no matching control, or plant in the scope that never reaches the hazard register, is found rather than assumed.

   What you see

   A progress screen with the stage it is on — uploading, extracting, analysing, generating. It is not a spinner pretending to be busy; each stage completes in turn.

   Why it matters

   Fatigue does not affect the tenth document differently from the first here. That is the entire proposition: the same standard applied at 4pm on a Friday as at 9am on a Monday.

3. 03

   The report

   Four severity levels, a section reference on every finding, and the reasoning behind each one.

   What you do

   Read the four counts first — Critical, Needs Attention, Minor and Done Well — and you know the shape of the document before you have read a single finding. Then go to Critical.

   What happens in the background

   Findings are sorted by severity, not by the order they were found. Each one is anchored to the section it came from and carries a "why it matters" you can open, so the reasoning is on the page rather than implied by the fact that something was flagged.

   What you see

   The report on screen, downloadable as a PDF, and as an editable Word file on Growth and above. You can also share a read-only link with the principal contractor without giving them an account.

   Why it matters

   A list of problems with no severity ranking makes everything look equally urgent, which in practice means nothing does. A flag with no reasoning attached cannot be disagreed with — and disagreeing with a flag is a legitimate outcome. Valify raises the question; your team makes the call.

4. 04

   The decision

   Accept the document, or return it for revision — and either way it is on the record.

   What you do

   Once your team has reviewed the findings, record your decision directly in Valify. Accept the document — logging who approved it, when and what the report contained at that moment — or return it for revision. Every decision goes into your audit trail and cannot be edited afterwards.

   What happens in the background

   The decision is written as an insert, never an update: there is no path from the browser that can edit or delete one afterwards. It carries the severity counts as they stood at the moment of the decision, snapshotted rather than joined, so the record reads correctly on its own. It is kept with the document and deleted with it at 90 days, like everything else about an upload — export it before then if you need it on file for longer.

   What you see

   The decision panel under the report, and afterwards a line on the document showing who decided, what they decided and when. The document history page can be filtered by accepted, returned or still pending.

   Why it matters

   When something goes wrong the question is who reviewed this document and what they concluded. A report on its own does not answer that; a report plus a named, timestamped decision does. Accepting in Valify records your decision — it is not a Valify approval, and nothing here moves that line.

5. 05

   Fixing

   The part only your people can do.

   What you do

   Work through Critical first. Each finding names the section, states what is missing and says what a thorough reviewer would have expected to find there — so the fix is usually a specific edit rather than a rewrite. Findings you disagree with, you close. That is a judgement call, and it is yours.

   What happens in the background

   Nothing. Valify does not edit your document, and it does not sign anything off. It has no view on whether a control measure is adequate for your site; it has a view on whether the document says what it needs to say.

   What you see

   The report stays in your document history, so what was flagged and what you did about it is on the record.

   Why it matters

   A tool that rewrote your RAMS would be producing a document nobody competent had authored, for a purpose where a competent person has to be accountable. The findings are the deliverable. The document stays yours.

6. 06

   Resubmit

   Run the corrected version through again and compare.

   What you do

   Upload the revised document as a resubmission of the original. It runs the same review.

   What happens in the background

   The new report is produced independently — it is a fresh review, not a diff of your edits — and then linked to the original so the two can be read side by side.

   What you see

   A comparison of the two reviews: what has been resolved, what is still open, and anything the revision introduced. Editing a document to close one finding occasionally opens another, and this is where that shows up.

   Why it matters

   A first report tells you what is wrong. The second tells you whether you fixed it — which is the thing an auditor is actually asking about when they ask what your review process is.

The cross-document checks, in plain English

Checking each section on its own is the easy half. These are the checks that read one section against another — the ones that are genuinely hard to do by hand on a forty-page document, because they require holding two distant pages in your head at the same time.

• Does every hazard have a control measure?

  Each entry in the hazard register is matched against the control measures section. A hazard listed in Section 6 with nothing in Section 8 that addresses it is flagged — as is a control measure that addresses a hazard nobody wrote down.

  How it gets missed: The two lists are usually pages apart and often in different formats — a table on one page, prose on another. Checking them against each other means holding fifteen or twenty hazards in your head while reading a different section.

• Does every item of plant appear in the hazard register?

  Plant and equipment named in the scope or the plant section is looked for in the hazard register. A cut-off saw in the equipment list with no entry for dust, noise or vibration anywhere is the classic version of this gap.

  How it gets missed: Plant is usually added late — a machine gets swapped in the week before the works — and the equipment list is updated while the hazard register, written first, is not.

• Does the PPE match the hazards, and follow the controls?

  The PPE table is read against the identified hazards two ways: is there PPE listed for each hazard that needs it, and is PPE being used to do a job an engineering control should have done first.

  How it gets missed: A PPE list looks complete on its own. It is only wrong in relation to something else — a hazard it does not cover, or a control measure it is quietly substituting for.

• Do the hazards match the scope of works?

  Every activity described in the scope and the method statement is checked for a corresponding hazard entry. If the sequence describes a Day 2 extension and the hazard register only covers initial erection, that is flagged.

  How it gets missed: Scope changes during drafting. The method statement gets the new activity, the hazard register keeps the old shape, and both sections read correctly on their own.

• Do the materials match the COSHH assessments?

  Substances named anywhere in the document are checked against the COSHH assessments, in both directions — a material with no assessment, and an assessment for something the works no longer involve.

  How it gets missed: COSHH assessments are frequently attached as a pack from a previous job. Reading them confirms they exist; it does not confirm they are for these materials.

• Is the document internally consistent about the basics?

  Site address, client, principal contractor, dates and project reference are compared everywhere they appear. So are the revision history and the sign-off dates — a document approved after the works were scheduled to start is a finding.

  How it gets missed: These are the details a reviewer reads past on the first page and never returns to, which is exactly why a stale muster point or a previous client’s name survives three revisions.