AI-Powered Reporting
AI reads the photo and the words of every report, points out what a trained eye would check first and drafts the next step. Your safety professionals review, decide and stay accountable.
Reading the photo
Suggested hazards
3 for review
Preliminary severity High
How AI understands observations
Workers report the way that suits the moment — a photo, a few words, a voice note. The AI turns what it receives into suggestions your register understands, in the language of the report.
AI suggestion
Validated against the values your register allows
What it suggests
AI is assistive. It does not replace trained professionals, formal risk assessments or approvals — and without an AI provider, reporting works exactly as before.
Interactive demonstrationSample data
Step through an AI photo analysis on sample data. At the review step, accept, edit or dismiss each suggestion — your choices decide what is saved.
AI photo analysis
Step 1 of 7: Photo received. A site photo arrives with a report from the web, the mobile app or WhatsApp.
Sample photo and sample results. In the product, AI analysis runs only where your organization has switched it on and an AI provider is configured.
Risk and action recommendations
Suggestions are written to be useful on site: specific to what is in the photo or the report, limited to the values your register allows, and always marked as suggestions.
Possible hazards with the reason they were flagged, so the reviewer knows where to look.
A level from your register’s severity scale as a starting point. The reviewer sets the final severity.
Practical first actions that can pre-fill a corrective action — owner and due date are still set by people.
A suggested observation type and a category chosen from your own list, never an invented one.
Suggestion for OBS-0212
AI suggestion · for reviewPreliminary severity
Suggested corrective action
Install barriers or edge protection along the trench and keep the walkway separated.
Use as a corrective action
Severity, not a risk score. The AI suggests a preliminary severity and the hazards behind it. It does not calculate a risk rating or replace your risk assessment process.
From data to management insight
Beyond single reports, AI helps people work with their safety data — always on records they are allowed to see.
Human oversight and governance
Every AI result is a draft until someone with the right permissions accepts it. The platform enforces that — it is not left to good intentions.
01
Suggestions, drafts and possible findings.
02
Accepts, edits or dismisses each one.
03
Only a confirmation creates or changes a record.
04
The saved record carries the reviewer’s choices.
Confirmation required
AI may prepare a record; a person must confirm it. The default.
Draft only
AI may only prepare drafts for people to complete.
Read only
AI may read and explain, but never prepare records.
Prohibited
AI is not used for that record type at all.
Privacy and data controls
AI is disabled by default. Your organization opts in and an AI provider is configured before anything is sent.
In Green Codes Pro, answers are built only from records the person could already open in full.
In Green Codes Pro, personal, medical, commercial, investigation and legal data go only to AI connections approved for them — otherwise the request is refused.
In Green Codes Cloud, reporter contact details are never sent to the AI provider.
What people type is wrapped and treated as untrusted data, guarding against attempts to steer the AI.
Each AI call is logged with feature, model, tokens and cost. Cloud stores no prompt or response text; Pro keeps a short excerpt, which can be turned off.
Supported AI providers
Green Codes Pro: Anthropic, OpenAI, Azure OpenAI and Google Gemini, assigned per AI feature.
Green Codes Cloud: Anthropic for analysis and an OpenAI-compatible service for voice transcription.
Business outcomes
Reviewers start from a structured suggestion instead of a blank form.
Types, severities and categories come from your own lists, report after report.
Professionals spend their time judging and acting, not re-typing.
Suggestions are written in the language of the report.
Every saved decision has a person behind it.
Availability by edition
AI features are available in every edition and run only where your organization has switched them on and an AI provider is configured.
AI analysis of each observation — summary, type, severity, category, hazards and recommended actions — applied by a reviewer; plus summaries, duplicate hints and voice transcription.
Photo suggestions, AI vision, drafting with confirmation, governance per record type and data-category controls across the main safety modules of Green Codes Pro.
Everything in Cloud Pro across every Green Codes Pro module, with AI providers, governance and data categories agreed for your organization.
Questions
AI raises good questions. Here are the ones we hear most.
Related features
Tell us how your teams report today, and we will prepare a quotation with the AI features that fit your edition.
