Survey clock18:15:14UK

4 AI tools live7 core services10 expertise areasISO 45001ISO 14001ISO 9001NEBOSHIOSHISEP

Advisory activeActive

Ask Precog AI · live now

AI Safety Consultantask the question. see the structure.Start with the real work

Explore a complex safety challenge through a guided, context-aware conversation with clear human-review boundaries.

Open the tool

What to bring

Ask Precog AI

00sections, every run

A focused conversation for organizing what you know and identifying what needs a competent human decision.

Things to have ready

Checked before you invest

Sections in every output

The same shape, every run

Things it will not do

Stated up front

Built around evidence

Tools

Guided safety conversation

Bring the context. We will help organize the questions.

The consultant is designed to make complex issues easier to discuss. It will ask for missing context, prioritise prevention and make uncertainty visible.

Precog AI safety consultant

Guided conversation · human-reviewed output

Start with the work

What safety challenge are you working through?

Share the activity, people involved, environment and controls you know about. The consultant will ask for what is missing.

AI output requires validation by a competent person before use.

Before you start

Bring these four things.

The tools are only as useful as the context you bring to them. Tick off what you already have to hand.

Context readiness

Nothing ticked yet.

Partial context still helps. It is more useful to start than to wait until you have everything.

The response

What comes back, in the order it arrives.

Structure, not a verdict. Every part below is a draft for a competent reviewer to accept, correct or reject.

Part 01

A structured summary

The challenge restated in your terms, so you can see whether it has been understood before you react to it.

Part 02

Hazards and exposure

What could go wrong, who might be affected, and where the conversation usually gets thin.

Part 03

Control options by hierarchy

Prevention, protection and administrative options, framed as options to evaluate rather than a chosen control.

Part 04

Open questions

What the tool could not resolve from what you gave it, and what to check before acting on the output.

Getting a useful result

How to work through it.

Three steps, in order. Skipping the second one is the most common reason a draft comes back thin.

01
Describe it plainly
Write the situation as you would brief a competent person, without trying to sound technical.
02
Answer the follow-ups
The assistant asks for the missing context that usually changes the answer. This is where the value is.
03
Take it to a review
Use the output as the agenda for a conversation with the people who will own the decision.

Design principles

Built to be argued with.

What makes this tool behave the way it does, stated plainly.

Ask better questions

Get a clear structure for a conversation with your team, client or competent reviewer.

Keep control

The tool supports analysis. It does not approve work, certify compliance or replace professional judgement.

Protect the context

Do not enter personal data or information that could identify an individual. Review provider settings before sensitive use.

The review boundary

What this tool will not do for you.

Stated here rather than buried, because a tool that overstates what it is will eventually be trusted at the exact moment it should not be.

How we approach responsible AI

It does not approve work, permit it, or certify that a task is safe.
It is not a substitute for a site-specific risk assessment.
It is not legal advice and cannot interpret a contract or a regulation for you.
It cannot know your site, your equipment, your history or your organization's tolerances.

Choose your starting point

One of these four fits. The other three do not.

All four tools are live. This is the honest version of which one to open, including when the answer is a different page.

You are here