THE KEYSTONE

The risk named.
The control held.

SafetrexIQ drafts your ISO 45001 management system, then keeps every risk, incident and corrective action on one auditable record. You stay the competent person.

Create your organisationBook a demo

NO CARD REQUIRED · EU DATA RESIDENCY · AI OFF BY CHOICE

ISO 45001

ALIGNED STRUCTURE

RIDDOR

DETERMINISTIC VERDICT

FR / EU

RECORDS STAY HERE

1 AFTERNOON

TO A DRAFTED SYSTEM

IN THE FIELD

Built hands-on with a chartered UK health and safety professional, and running in live pilot with a UK logistics operator.

JURISDICTIONUNITED KINGDOMSTATUTE COVERAGERIDDOR · COSHH · HASAWATENANCYISOLATED PER ORGSTATUSMVP · PILOT LIVE

HOW IT WORKS

01

Describe the company once

Sector, sites, headcount, the activities you actually do. Ten minutes, editable forever.

PROFILE → JURISDICTION PACK
UK SELECTED BY DEFAULT

02

Your system is drafted, cited

Statement of intent, 20 policies, roles and RACI, a scored 5×5 register — every claim traceable to a clause or a statute.

DRAFT → REVIEWED → APPROVED
NAMED APPROVER ON RECORD

03

Then it runs the loop

Incidents investigated, reportability judged, actions owned and verified. Recurrence flagged when a control fails twice.

LIVE KPIs · OVERDUE ALERTS
EVIDENCE FOR THE AUDITOR

THE INSTRUMENTS

Three things an auditor asks to see

FIG.1 — 5×5 MATRIXEXTREME 25

Likelihood against severity, scored per activity. The cell values and the escalation threshold are fixed in code, not improvised by a model.

FIG.2 — THE LOOPCONTINUOUS
RISK ASSESSEDINCIDENT LOGGEDINVESTIGATEDRIDDOR VERDICTACTION CLOSEDCONTROL VERIFIED

Every incident returns to the register: the control that failed is the control you reassess.

FIG.3 — REPORTABILITYADVISORY

Reportable

RIDDOR REG.6(1)

TRIGGEROVER-7-DAY ABSENCEDEADLINE15 DAYSDECIDED BYRULES, NOT AI

Advisory only. SafetrexIQ does not file the report — it tells you the duty exists, in time to meet it.

MODULE DIRECTORY

ALL 24 MODULES →

01 / SYS.GENERATOR

Your management system,
drafted in an afternoon

Statement of intent, a 20-policy suite, roles and RACI with legal duties, and a scored risk register — cited, and yours to approve.

DRAFTREVIEWEDAPPROVED

02

Incident investigation

Contributing factors mapped to the control that failed.

03

Corrective actions

Owned, dated, verified for effectiveness.

04

Audits & inspections

Findings on a phone, actions raised in the same step.

05

Anonymous intake

QR reporting for any worker. No login.

06 / ALERT.BUS

RIDDOR checks and overdue alerts

A duty or a deadline never quietly slips.

07

Competency & training

Who is qualified, and when it expires.

08

Contractors & permits

Approval state before anyone sets foot on site.

AI PRIVACY

The AI never
learns who.

Records are stored in France, inside the EU. Every request is classified before it runs: health information is held back and names are replaced with placeholders, so the model reasons about the safety problem without being told whose problem it is. You choose how much it may use, and you can switch it off entirely.

READ THE ARCHITECTURE →
AI GATEWAY — EVERY REQUEST, EVERY FIELD
1ClassifyPER FIELD
2MinimiseLEAST DATA
3Tokenise namesPERSON_A
4Hold back health dataON-PREM ONLY
5Decide by your policyOR REFUSE

RIDDOR, PLAINLY

Know the duty before
the deadline.

RIDDOR makes certain events reportable to the HSE within fixed windows. SafetrexIQ judges reportability from the facts you log, with the regulation named, and tells you what it cannot decide for you.

Specified injuryWITHOUT DELAY · 10 DAYS
Over-7-day absence15 DAYS
Dangerous occurrenceWITHOUT DELAY
Near miss, no injuryRECORD ONLY

Advisory. Reporting remains your duty, and a competent person confirms the verdict.

INTEGRATIONS

It fits the tools
already on site.

EMAIL & CALENDAR

Alerts and due dates where people already look

SSO

One sign-in, roles enforced server-side

PDF EXPORT

Any document, audit-ready, versioned

QR / LINK INTAKE

Printed on the wall, no app to install

ON-PREM MODEL

Health data never leaves your path

API

On the roadmap

PRICING

COMPARE PLANS →

SOLE TRADER

One site,
one record

The drafted system and the risk register, for a single site.

SME

The whole
loop, running

Unlimited sites, RIDDOR checks, AI drafting with citations.

ENTERPRISE

On your own
infrastructure

On-prem model path, SSO, OH data separation.

Request a demo

We scope the plan on the call — sites, headcount and how much of the AI you want on.

FAQ

The questions an auditor asks first.

Does this certify us to ISO 45001?+

No. Certification follows an independent audit by an accredited body. SafetrexIQ builds and maintains the management system that audit examines.

Is AI-generated content safe to rely on?+

Everything arrives as a draft with citations, and moves to approved only when a competent person signs it off. The approver and the version stay on the record.

Where does our data live?+

In France, inside the EU. Health and occupational-health records sit in a separate store behind an OH-role gate, with longer retention.

Does it file RIDDOR reports for us?+

No. It gives a deterministic reportability verdict with the regulation cited and the deadline stated. Submitting to the HSE remains your duty.

Can another company see our records?+

No. Every table carries an organisation id and every query filters on the signed-in session's organisation and role, enforced on the server.

DEPLOYMENT

Stand up your safety
system this afternoon.

Create your organisation, answer a short profile, and SafetrexIQ drafts the system for your review. You stay the competent person; it does the assembly and the follow-through.

Create your organisationBook a demo

Generated frameworks, policies and risk assessments are drafts for a competent person to review and approve. Nothing here confers ISO 45001 certification, which follows an independent audit by an accredited body. Jurisdiction support is United Kingdom first.