Featherstone Safety HubLog inStart free trialFree trial
Featherstone Safety

Featherstone AI

AI-assisted health and safety management

Featherstone AI is an assistant built into Featherstone Safety Hub. It drafts risk assessments, COSHH assessments and RAMS from a short description, and tells you in plain English what is overdue, due soon or needs following up across your own records. It is there to cut the admin, not the judgement: every draft is a starting point for a competent person to check, edit and approve.

Included on every plan with a monthly usage allowance per organisation. It stays off until an admin of your organisation switches it on, and an admin can switch it off at any time.

Written by Thomas Featherstone TechIOSH, NEBOSH IGC and NEBOSH Environmental Management Certificate qualified, and IOSH Managing Safely trained · Reviewed June 2026

No long-term contractsCancel any time.
Transparent pricingFrom £49 per month.
Built by a NEBOSH-qualified practitionerReal H&S expertise built in.

What it does, and what it does not

Most of the time spent on health and safety paperwork goes on two things: starting documents from a blank page, and working out what has slipped. Featherstone AI helps with both, inside the same system that holds your assessments, training, actions and incidents.

It does not sign anything off, approve anything or decide that you are compliant. Drafts open in the normal forms as drafts, and nothing is saved until a person reviews it and presses Save. Using AI does not make a document suitable and sufficient: that still depends on someone who knows the work checking it against what actually happens on site.

Draft a risk assessment from a short description

Describe the task, for example "operating a mobile elevating work platform in a warehouse", and Featherstone AI returns a draft in the Hub's own risk assessment form: the significant hazards, the harm and who could be harmed, recommended controls following the hierarchy of control, further actions, and initial and residual scores.

The scores are recalculated with the Hub's own 5×5 matrix, not taken on trust from the model. Where your organisation already has assessments for similar work, the draft reuses wording from up to two of them, so it reads like your documents rather than a generic template.

  • Recommendations, not assumptions: controls are written as recommendations to confirm. The draft never claims a control is already in place.
  • No invented facts: it does not make up site conditions, equipment, inspections or anyone's competence, and it lists what you need to confirm.
  • Your form, your sign-off: the draft fills the normal form as a Draft. You edit it, save it and set its review date as usual.

COSHH drafts that keep the safety data sheet in charge

Facts that belong to a specific substance, such as hazard and precautionary statements, CLP classification, pictograms and workplace exposure limits, must come from the supplier's safety data sheet. A COSHH draft from Featherstone AI never supplies them. It uses only what you copy in from the safety data sheet, and the Hub screens the draft and removes any sentence it finds naming such a fact that you did not provide. Your own check against the safety data sheet remains the final safeguard.

The general sections (routes of exposure, storage, first aid, spills, disposal and monitoring) are drafted for you and labelled with the section of the safety data sheet to check them against.

RAMS drafts from your own procedures (Construction plan)

On the Construction plan, Featherstone AI drafts RAMS from a description of the job, drawing on your own method statement procedures, PPE items, equipment and risk assessments. It can only point to items in your libraries, and their content is copied from your records rather than written by the model.

Site facts such as addresses, dates, people, contacts and permits are left blank for the author. A RAMS with AI content cannot be issued until those details are filled in and the AI content has been checked.

See what needs attention, in plain English

The workload summary is worked out by Featherstone Safety Hub itself from fixed rules over your records' dates and statuses: overdue corrective actions, assessment reviews, expiring training, fire checks, contractor documents and more. It works whether or not AI is switched on, and it keeps working if the monthly allowance is used up.

Ask a simple question such as "any overdue actions?" and it is answered directly from your records, with no AI call. Questions that need judgement, such as "what should we focus on this week?", go to the assistant, which explains and prioritises what the rules found. The assistant never decides for itself what is overdue.

  • Confirmed and to check: findings where a date has passed are kept apart from possible gaps that a competent person should look at.
  • Linked to the record: each finding links to the record the Hub actually read, so you can go straight to it.

How your data is handled

While Featherstone AI is off, nothing is sent to an AI provider. When it is on, the Hub sends our provider, Anthropic, only what is needed for the answer or draft: what the user typed and the relevant record details such as titles, dates, statuses and categories. Names attached to the records being sent are removed, and so are your employees' names in drafts. Files are not sent.

Anthropic is US-incorporated and may process the data outside the UK, under the Standard Contractual Clauses and the UK International Data Transfer Addendum. Its published terms say API data is not used to train its models by default and is deleted within 30 days. We do not store the questions or answers, only usage records for up to 13 months. The full detail is in section 4A of our privacy policy and on the subprocessor list.

See it in the app

Inside Featherstone Safety Hub

Real screens from the live app, not mock-ups.

Risk assessment form in Featherstone Safety Hub, the form an AI draft opens in
An AI risk assessment draft opens in this normal form, marked as a Draft. Nothing is saved until you review it and press Save.
COSHH assessment form in Featherstone Safety Hub showing the hazard statement field copied from the safety data sheet
COSHH drafts never supply hazard statements, pictograms or exposure limits. Those come only from what you copy in from the safety data sheet.

Free, no sign-up

Not sure where your business stands?

Take the free health and safety compliance check. Answer nine quick questions and get an instant, personalised report of indicative compliance gaps based on your answers, the regulations behind them and how to address them. Not legal advice. No email required.

Start the free check →

Explore Featherstone Safety Hub

Related solutions

Health and safety management software →Risk Assessment Software →COSHH Assessment Software →RAMS Builder Software →Action Tracking Software →For Small Business →

Related guides

Free health and safety guides

How to do a risk assessment: the 5 steps with a worked example →How Featherstone AI handles your data →
A UK company
Featherstone Safety Ltd, registered in England and Wales, company number 17410006.
Built by a safety practitioner
Thomas Featherstone TechIOSH, NEBOSH IGC and NEBOSH Environmental Management Certificate qualified, and IOSH Managing Safely trained.
Data protection
ICO registration 00013953517. Data hosted in Ireland (EEA), exportable at any time. See security and data protection and the privacy policy.
Real support
Email thomas@featherstonesafetyhub.co.uk or call 01865 570810.

Frequently asked questions

Common questions

What is AI health and safety software?

It is health and safety software with built-in AI assistance. In Featherstone Safety Hub, the AI drafts risk assessments, COSHH assessments and RAMS for a competent person to review, and explains in plain English what is overdue or due soon across your records.

Can AI write my risk assessments for me?

It can write a first draft, but not a finished assessment. The law requires a suitable and sufficient assessment, and that depends on someone who knows the work checking the draft against real conditions, adding what only they know and approving it. Featherstone AI drafts are opened as drafts for exactly that reason.

Does using AI make my documents compliant?

No. A draft from Featherstone AI is a starting point, not professional advice and not evidence of compliance on its own. Compliance comes from the assessment being right for your work, the controls being in place and the document being reviewed.

Where do COSHH drafts get hazard information from?

Only from the safety data sheet. Hazard and precautionary statements, pictograms and exposure limits are never supplied by the AI. You copy them in from the supplier's safety data sheet, and the Hub screens the draft for any such fact you did not provide and removes it. You should still check the finished assessment against the safety data sheet.

Is Featherstone AI included in the price?

Yes. It is included on every plan with a monthly usage allowance per organisation. If the allowance is used up, AI answers and drafts pause until the 1st of the next month, and the workload summary keeps working. Plans start at £49 per month with a 14-day free trial.

Is my data used to train AI models?

Our AI provider, Anthropic, says in its published terms that API inputs and outputs are not used to train its models by default, and are deleted within 30 days except where needed to enforce its usage policy. We send only what is needed for each answer, remove the names attached to the records sent, and do not store the questions or answers.

Does my data leave the UK?

When Featherstone AI is switched on, the details sent for an answer or draft may be processed by Anthropic outside the UK, under the Standard Contractual Clauses and the UK International Data Transfer Addendum. While it is off, nothing is sent to an AI provider. Our privacy policy lists where every part of the service is hosted.

Can we switch it off?

Yes. It is off until an admin of your organisation switches it on, and an admin can switch it off at any time.

Start from a draft, not a blank page

Start a free 14-day trial, switch on Featherstone AI and draft your first risk assessment, ready for you to review and approve. No setup fee, cancel any time.