AI tools that support your professional judgement, not replace it

There's a lot of noise around AI in health and safety. Some tools claim to write assessments for you. The problem with that framing is that a risk assessment is a professional judgement about real conditions at a specific site, and that judgement can't be outsourced to software. What AI can usefully do is remove the blank page, generate a considered starting point and let the qualified assessor get straight to reviewing rather than drafting.

AI drafts, you review and approve Professional judgement stays with you Optional, start from scratch if you prefer

What the AI draft assistant does

When you start a new assessment in Lytom Assess, the optional AI draft assistant can generate a structured starting point based on the assessment type you have selected. For a fire risk assessment, it will produce a set of hazard categories relevant to the building type, a suggested approach to each area of consideration, and a list of common control measures for review. For a COSHH assessment, it will suggest an initial substance classification, exposure route considerations and typical control hierarchy.

Every item it produces is editable. The draft is a scaffold, not a finished document. The qualified consultant adjusts it to reflect what they actually found on site, removes irrelevant items, adds site-specific findings and applies their professional view of likelihood and severity before issuing anything to a client.

What AI should and should not do in H&S assessments

Why this matters for your professional liability

A risk assessment that goes out under your name carries your professional liability. If an AI tool writes something and you issue it uncritically, the professional and legal exposure doesn't sit with the software company. It sits with you. Lytom Assess is designed so that the consultant remains the author. The AI produces a draft that you review, amend and approve. The final document reflects your assessment of the site, informed by your experience, your site visit and your professional training.

Assessment types where the draft assistant is available

• General risk assessment

  Workplace activity-based general risk assessments, adaptable to most common workplace hazard categories.

• Fire risk assessment

  Structured fire risk assessment draft covering the main areas of consideration under the Regulatory Reform (Fire Safety) Order 2005.

• COSHH assessment

  Substance-level COSHH assessment draft with suggested hazard classification, exposure routes and control hierarchy for common substance types.

• Working at height

  Draft covering hierarchy of controls, common equipment types and typical risk factors for work at height activities.

Frequently asked questions

Does the AI write the risk assessment for me?

No. The AI drafts a starting point: hazards, suggested controls and structure. You review every item, adjust for the site, apply your professional judgement and issue. The assessor remains responsible for the content.

What does the AI draft assistant actually produce?

Given an assessment type and context, it generates suggested hazard categories with initial ratings, a list of relevant control measures, and a structured starting template. A considered starting point rather than a blank page.

Is AI-assisted output legally sound and compliant?

It's a draft only. Legal compliance depends on the assessor's review and the accuracy of the final document. The qualified consultant is responsible for verifying all site-specific hazards and checking controls are appropriate and achievable before issuing.

Can I turn off the AI draft feature?

Yes. It is an optional feature. You can start any assessment from a blank template or from your own saved template if you prefer your own starting framework.

Does the AI have access to my client data?

The draft assistant uses only the context you provide: assessment type and any notes you enter. It doesn't access other client records or stored data without your instruction.