Skip to content

Automatic Hazard Identification

AI hazard identification analyses your document and suggests hazards you may have missed. It also recommends control measures for specific hazards following the hierarchy of control.

Use this when you have written your work scope and want to check for gaps in your risk assessment.

  1. Open your document in the editor.
  2. Go to the Risk Assessment section.
  3. Click Suggest Hazards (or Find Missing Hazards).
  4. AI analyses your work description, method statement, equipment list, and site details.
  5. A panel appears with suggested hazards.

[Screenshot: Hazard suggestion panel showing AI-identified hazards with confidence scores and add buttons]

Each suggested hazard includes:

  • Hazard name — a clear description of the hazard
  • Category — e.g. work at height, electrical, manual handling, noise
  • Risk description — how the hazard could cause harm
  • Confidence score — how confident AI is that this hazard applies to your work (high, medium, or low)
  • Suggested severity and likelihood — initial risk scores
  • Suggested control measures — with references to relevant UK regulations

For each suggested hazard, you can:

  • Add — adds the hazard to your risk assessment. You can then edit the scores and controls.
  • Dismiss — removes the suggestion. Dismissed hazards do not appear in future suggestions for this document.
  • Edit and add — modify the hazard details before adding it.

Cost: 2 credits per suggestion run.

  • Complete your work scope first — AI uses this to understand what you are doing. A vague scope produces vague hazards.
  • List your equipment — AI identifies equipment-specific hazards (e.g. MEWP → falling objects, battery failure; angle grinder → hot works, vibration)
  • Include site details — mention overhead services, confined spaces, nearby roads, asbestos, or anything unusual about the site
  • Run it after editing — if you significantly change the work scope or method statement, run hazard identification again to check for new hazards

For any individual hazard in your risk assessment, AI can suggest additional or improved control measures.

  1. In the Risk Assessment section, click on a specific hazard.
  2. Click Suggest Controls.
  3. AI returns a list of recommended control measures.

[Screenshot: Control measure suggestions for a specific hazard, showing hierarchy levels]

AI organises suggestions following the hierarchy of control — the legally required approach under UK regulations:

LevelDescriptionExample
1. EliminationRemove the hazard entirelyUse pre-fabricated components instead of cutting on site
2. SubstitutionReplace with something less hazardousUse water-based paint instead of solvent-based
3. Engineering controlsPhysical measures to reduce exposureInstall edge protection, use LEV extraction
4. Administrative controlsProcedures and trainingPermit to work, toolbox talk, supervision
5. PPEPersonal protective equipmentHard hat, safety glasses, respiratory protection

AI indicates which hierarchy level each suggested control addresses. Higher levels are preferred — you should aim to eliminate or substitute where possible before relying on PPE.

For each suggestion, you can:

  • Add — adds the control measure to the hazard
  • Dismiss — removes the suggestion
  • Edit and add — modify the wording before adding

Cost: 1 credit per hazard.

When to use hazard identification vs. full generation

Section titled “When to use hazard identification vs. full generation”
ScenarioUse
Starting a new document from scratchFull document generation (5 credits) — generates everything including hazards
You have a document but want to check for missing hazardsHazard identification (2 credits)
You want better controls for a specific hazardControl measure suggestions (1 credit)
You are updating an existing document after conditions changedHazard identification (2 credits) — finds new hazards based on changes
  • AI identifies hazards based on your description. If you do not mention an aspect of the work, AI may miss related hazards.
  • Confidence scores are a guide, not a guarantee. A low-confidence suggestion may still be relevant to your work.
  • AI cannot assess site-specific conditions it cannot see. Upload site photos and describe unusual conditions to improve accuracy.
  • Always cross-check AI suggestions against your hazard library, your experience, and current site conditions.