Accident & Incident Reporting Software UK

Every incident captured, investigated and RIDDOR-ready.

Record accidents, near misses, dangerous occurrences and ill health with every injured person on one incident, photograph a handwritten form and let AI fill it in, track RIDDOR deadlines, run the investigation, and let the AI Health & Safety Manager surface the patterns and repeat offenders before they become the next one. Encrypted, scoped and aligned to RIDDOR 2013.

Person	Type	Body area	Severity	Hosp.	RIDDOR
Anna Williams Cellar staff · 12 days off	Employee	Hands · Left wrist	S4	✓ A&E	Pending
Tom Reilly Cellar staff · 4 days off	Employee	Legs · Right shin	S3	No	Not required
K. Hutchinson Bar customer	Public	Head · Eyes	S2	✓ A&E	Pending

Person

Type

Body area

Severity

Hosp.

RIDDOR

Anna Williams

Cellar staff · 12 days off

Employee

Hands · Left wrist

✓ A&E

Pending

Tom Reilly

Cellar staff · 4 days off

Employee

Legs · Right shin

Not required

K. Hutchinson

Bar customer

Public

Head · Eyes

✓ A&E

Pending

Eight problems paper accident & incident reporting keep causing.

Each one costs UK operators hours, ratings, insurance excess or worse. Each one disappears when the accident & incident reporting becomes digital.

The accident book is a paper pad in a drawer, and half of what happens never reaches it.

Cost:Near misses go unrecorded, the one incident that mattered is illegible or missing, and there's no defensible record when the HSE or your insurer asks.

Fixed:Log incidents in seconds - or photograph the handwritten accident-book page and let AI extract the details and pre-fill the form. The original form is attached to the record, and anyone can report anonymously through a public link without logging in.

A real incident injures more than one person, but the form only has room for one.

Cost:A scald that catches two staff, a slip that takes down a customer and an employee - recorded as one person, or as duplicate records nobody can reconcile, with the RIDDOR picture wrong from the start.

Fixed:One incident, every injured person on it. Each person carries their own injuries, body areas, severity, hospital visit, treatment, lost-time and RIDDOR status - captured, reported and exported per person, not flattened into one box.

RIDDOR deadlines are counted by hand, if at all, and the clock has usually already run out.

Cost:A late or missed RIDDOR report is a criminal offence and the first thing an inspector checks - and "we didn't realise it was reportable" is not a defence.

Fixed:Every injured person gets a RIDDOR status and a calculated deadline from the incident date. Unreported reportable incidents are escalated as critical in the AI Health & Safety Manager's daily brief, with the submission and the HSE acknowledgement recorded against the person.

The same thing keeps happening and nobody connects the dots until the serious one.

Cost:Three Monday-morning manual-handling injuries on deliveries, a cluster of late-night bar slips, one employee in the records again and again - invisible until the incident that triggers an investigation.

Fixed:The AI Health & Safety Manager reads ninety days of incidents and names the pattern in plain English - the hot-spot zone, the time-of-day cluster, the repeat offender - in the leadership daily brief, before it becomes the next claim.

"We investigated it" means a note somewhere and an action nobody owns.

Cost:No status, no owner, no due date, no proof the corrective action was ever done - exactly what fails an audit and what a personal-injury solicitor goes looking for.

Fixed:A customisable investigation workflow with status history, an investigation team with roles, and corrective and preventive actions that carry an owner, a priority, a due date and a verified-complete state - every status change logged, append-only.

Witness accounts are scribbled on loose paper, collected days late, or never taken at all.

Cost:The recollection that would have settled a liability claim is gone, or undated and unsigned, by the time anyone needs it.

Fixed:Capture witness statements in-app, or issue a single-use, time-limited public witness link the witness completes themselves with email verification - the account is taken while it's fresh, attributed and timestamped.

You can't prove what the record said, or who changed it, after the fact.

Cost:Months later a record has been edited, a person quietly removed, and the version that mattered on the day of the claim is a guess.

Fixed:Every field change on the incident and on each injured person is logged - old value, new value, who, when - in an append-only audit trail, with soft-delete retention so a RIDDOR record is never truly gone.

The personal and medical detail in an incident record is exactly what you can least afford to leak.

Cost:Names, addresses, injuries and treatment sitting in a spreadsheet or a shared drive is a data-protection incident waiting to become its own headline.

Fixed:All personal and medical data is encrypted at rest, searchable without being decrypted, and isolated per business - the sensitive record is protected by design, not by where you hid the file.

The defensible way for a UK business to record, investigate and report incidents.

Capture every accident, near miss, dangerous occurrence and case of ill health - with every injured person on one record - from a fast form, a photographed accident-book page, or an anonymous public link. Track RIDDOR deadlines, run a real investigation with owned corrective actions, take witness statements while they're fresh, explore the whole picture in a five-tab analytics dashboard, and let the AI Health & Safety Manager surface the trends, hot spots and repeat offenders in the leadership daily brief. Encrypted, audit-trailed and aligned to RIDDOR 2013.

From accident-book page to RIDDOR-ready in four steps.

STEP 01

Capture it fast

Log it in seconds, photograph the handwritten form for AI to extract, or take an anonymous public report - adding every injured person on the incident.

STEP 02

Triage and assign

Set classification and severity, flag for review, assign a lead and an investigation team, and let RIDDOR deadlines calculate per person.

STEP 03

Investigate and act

Move it through your status workflow, take witness statements, log external bodies, and raise corrective actions with owners and due dates.

STEP 04

Report and learn

Record the RIDDOR submission and HSE acknowledgement, export the branded PDF, and let the AI brief surface the pattern before it repeats.

Why digital incident reporting wins.

The paper accident book

Near misses and the illegible page that mattered never recorded.
One person per form, so multi-casualty incidents are wrong from the start.
RIDDOR deadlines counted by hand, usually after they've passed.
The same incident repeating, invisible until the serious one.
Investigations with no owner, no due date and no proof of action.
Personal and medical detail in a spreadsheet anyone can open.

AssistantManagerHQ incident reporting

Fast logging, AI form extraction, and anonymous public reporting.
Every injured person on one incident, reported and exported per person.
RIDDOR deadlines calculated and escalated before they lapse.
AI pattern analysis names the trend in the leadership brief.
Owned, dated, verified corrective actions with full status history.
Personal and medical data encrypted at rest and isolated per business.

Powerful features, simple to use.

Multiple injured persons

Every person on one incident, each with their own injuries, severity and RIDDOR status.

AI form extraction

Photograph a handwritten accident-book page; AI extracts and pre-fills the incident.

Anonymous reporting

A public link anyone can use without logging in, with the original form attached.

RIDDOR deadline tracking

A calculated reporting deadline and status per injured person, from the incident date.

AI pattern analysis

Ninety-day trends, hot-spot zones, time clusters and repeat offenders in the daily brief.

Investigation workflow

Customisable per-business statuses with an append-only status-change audit log.

Corrective actions

Owned, prioritised, dated CAPAs with a proposed-to-verified lifecycle.

Witness statements

Capture in-app or via a single-use, email-verified public witness link.

Analytics dashboard

Five tabs - overview, demographics, location, injuries and period-over-period trends.

Lost-time tracking

Days away, restricted duty, return-to-work, OH referral and fit-note upload per person.

Encrypted by design

Personal and medical data encrypted at rest, searchable and isolated per business.

Branded PDF & CSV export

Full-incident and single-person PDFs, plus filtered CSV for the board pack.

Every capability, organised by how you'll use them.

Skim the themes on the left, or scroll the full set. Each capability surfaces the specific frustration it removes and a real-world use case.

Get up and running fast

Photograph the accident book - AI fills the form in

The fastest record is the one nobody has to type. Photograph or upload the handwritten accident-book page or incident form - a phone snap, a scan, a PDF - and AI reads the handwriting and pre-fills the incident and the injured-person fields for you to check and confirm. The original form is attached to the record and flagged as the source document, so the paper and the digital record never drift apart. The "I'll write it up later" gap that loses incidents disappears.

Use caseA kitchen porter snaps the accident-book page on his phone; the form fills itself in, the manager checks it, and the RIDDOR clock starts the same shift.

Anyone can report it - even anonymously

Not everyone who sees an incident has a login, and not everyone will put their name to it. A public reporting link lets anyone submit an incident without an account, anonymously if they choose, with the submission attributed and retained for the managers who need to act on it. The near miss that would never have been raised through a manager gets raised.

Use caseA contractor reports a near-miss with a falling pallet through the public link after hours; it's waiting for the duty manager in the morning, no login required.

Every injured person on one incident

A scald that catches two staff, a slip that takes down a customer and the employee who went to help - one event, every person on it. Add as many injured persons as the incident had, each with their own person type (employee, customer, contractor, supplier, member of the public), identity, injuries, affected body areas, severity, hospital visit, treatment given and first aider. The multi-casualty incident is recorded correctly from the first save, not reconstructed from duplicates later.

Fast, structured capture

When you do type it, the form is built for speed and for the regulator: date and time, classification - accident, near miss, dangerous occurrence, ill health or property damage - actual and potential severity, what happened, equipment involved, contributing factors, the zones and areas it happened in, and custom colour-coded categories you define. Flag it for review with a reason as you save, so the serious one is escalated immediately.

Built for service businesses

RIDDOR deadlines, tracked per person

RIDDOR reportability is a per-person question, so it's tracked per person. Each injured person carries a RIDDOR status and a reporting deadline calculated from the incident date - the shorter deadline for specified injuries and hospital cases, the longer one for over-seven-day lost time. The platform tracks the obligation and records the submission reference, the date, who reported it and what was reported, plus the HSE acknowledgement when it comes back. It manages the deadline and the evidence; you file with HSE.

Use caseA fracture at a soft-play site is logged; the ten-day specified-injury deadline is set automatically and the unreported flag clears only once the F2508 reference is recorded against the person.

Lost-time and return-to-work

For an injured employee, the record carries the full absence picture: days away from work, days on restricted duties, return-to-work date, expected return, still-off-work status, restricted-duty and phased-return notes, occupational-health referral and the fit note itself, uploaded to the record. The over-seven-day RIDDOR trigger and the cost of the incident both come from real data, not an estimate.

Severity, classification and the body-area map

Actual versus potential severity on a one-to-five scale captures the near miss that could have been a fatality, not just the cut that happened. Each injured person's injuries are mapped to specific body areas from a structured taxonomy - head, neck, arms, hands, torso, legs, feet - so the analytics can show where people are getting hurt, not just how often.

Custom fields and categories for your operation

Define the incident categories your business actually uses, colour-coded, and add the custom fields your sector or your insurer asks for. The record matches how you classify incidents, instead of forcing your operation into a fixed form.

Encrypted, searchable, isolated

Names, addresses, contact details, injuries and treatment are encrypted at rest and remain searchable without being decrypted, so you can find a person's record without exposing the data. Every record is isolated to your business. The most sensitive data in the platform is protected by design, which is exactly what the ICO and your insurer expect to see.

Real-time alerts

The AI Health & Safety Manager reads the trend, not just the count

Every day, the AI Health & Safety Manager reads the last ninety days of incidents and reports the pattern in plain English in the leadership brief - the hot-spot zone, the time-of-day cluster (the late-night bar slips, the Monday-morning delivery strains), the recurring body area, the classification creeping up month over month against the same month last year. It names the repeat offenders - staff with two or more incidents in the window - and escalates unreported RIDDOR as critical, with a deep link straight to the action. The pattern is told to leadership before it becomes the next claim, not found in a spreadsheet after.

Use caseThe brief opens with "three manual-handling injuries on the early delivery shift in the last fortnight, same zone" and a one-click jump to those incidents - the rota and the lifting brief change before the fourth.

Severity-tiered notifications

Notifications are tiered by severity and event, so the right people hear about the right things: the assignee, watchers and the creator on status changes and actions; venue admins escalated in on high-severity incidents, RIDDOR thresholds and SLA breaches. A serious incident isn't waiting in a queue for someone to notice it.

SLA tracking on every incident

Each incident gets an investigation SLA based on its severity - tighter for the serious ones. The platform tracks it, shows the due date and a breach warning on the incident, and stops chasing once the incident reaches a closed state. Investigations don't quietly age past the point a regulator would expect them closed.

Flag, watch and know it was seen

Flag an incident for review with a reason and it's escalated immediately. Add watchers who should be kept informed without being assigned. And because the platform records when the assigned investigator actually opened the incident, "assigned" and "looked at" stop being the same word - the AI brief calls out the assigned incidents the owner hasn't read.

Cross-module integrations

Scoped to the same estate as everything else

Incidents are scoped to the same zones and areas as Risk Assessments, COSHH and Checklists, with the same group and role-based access. An incident in a zone sits in the same operational geography as the risk assessment that should have controlled it and the checks that were meant to carry that control out - one estate, one model, not a separate accident silo.

On the operations dashboard, next to everything else

An incident widget on the main operations dashboard puts the period counts and the recent trend in front of the responsible managers every time they log in - alongside Checklists, Risk Assessments and COSHH, scoped to the zones and roles they own. Safety performance is one picture, not a report someone has to go and pull.

Feeds the leadership safety brief

Incident trends, repeat offenders and unreported RIDDOR sit in the same AI Health & Safety Manager daily brief as overdue risk assessments and COSHH reviews - one plain-English read of where the safety risk actually is across the business, with a deep link into each, instead of four dashboards nobody opens.

Connected to the HR absence and Bradford picture

When an injured person is an employee, the lost-time the incident caused is the same absence HR is managing - so an injury that puts someone off work flows toward the employee's absence and Bradford Factor picture in HR, rather than being counted twice or not at all. The incident, the absence it caused and the cost of it become one connected record across the platform.

Audit-ready reporting

A real investigation workflow

Move an incident through a status workflow you define - the default runs Open, In Review, Pending, Further Action, On Hold and two closed states, and you can change it per business. Every status change is written to an append-only transition log with a reason and a timestamp. Build an investigation team with roles - lead, manager, health & safety manager - kept in history even after someone is removed. "We investigated it" becomes a documented, attributable trail.

Corrective and preventive actions that get done

Corrective actions are first-class records, not a note. Each has a description, an owner, a priority, a due date and a lifecycle - proposed, approved, in progress, completed, verified - with who completed and who verified it, and when. The action that prevents the recurrence has an owner and a deadline and a proof-of-done, which is exactly what an auditor and a solicitor both ask for.

Witness statements, taken while they're fresh

Capture a witness statement in the app, or issue a single-use public witness link the witness completes themselves - time-limited, email-verified and anti-abuse protected - so the account is on record while the memory is accurate and clearly attributed, not a scribble collected a week later. External-body involvement (HSE, police, insurers) is tracked with a correspondence log alongside it.

Field-level audit trail and retention

Every change to the incident and to each injured person is logged - the field, the old value, the new value, who changed it and when - visible on the record's activity history. Deletions are soft, so a RIDDOR-relevant record is retained, not lost. When a claim or an inspection looks at this incident in two years, the record of what it said and who changed it is still there.

The incident analytics dashboard

A collapsible analytics dashboard sits on the incident list and remembers whether you keep it open. Four headline cards - total, last 7 days, last 30 days and last 90 days, each showing the movement against the previous period - sit on top of an Overview tab that breaks incidents down by hour of day, by weekday and by quarter. The shape of when incidents happen is a glance, not a query someone has to build.

Use caseA multi-site manager opens the dashboard, sees the Saturday-afternoon spike on the weekday chart, and moves a supervisor onto that shift.

Demographics, location and injury analytics

Three more tabs turn the records into the operational picture. Demographics breaks incidents down by age, person type and gender. Location ranks the zones and areas where incidents happen and shows average severity by zone, so the worst place is obvious, not anecdotal. Injuries shows the severity histogram on a green-to-red scale, the hospital-visit split and the body areas most affected - where people are getting hurt, not just how often.

Period comparison - the trend, week, month or year

The period-comparison tab tracks the metrics that actually matter as trend lines you can switch between weekly, monthly and yearly granularity: incidents with a known area, average severity, RIDDOR-needing persons and RIDDOR percentage, hospital visits and hospital percentage, people injured, and average age. Whether the safety picture is genuinely improving or just quieter this month becomes a line on a chart, not an argument in a meeting.

Use caseLeadership reviews the RIDDOR-percentage trend line quarter on quarter in the board pack instead of recounting incidents by hand.

Branded PDF and CSV, per incident or per person

Export a full incident as a branded PDF with every injured person and every detail, a single injured person as their own PDF for a RIDDOR pack or a claim, and the filtered incident list or the people-injured report as CSV for the board pack, the insurer or the audit. The evidence is a generated export, not a folder hunt.

Anna Williams	S4	Pending
Tom Reilly	S3	Not required
K. Hutchinson	S2	Pending