AI built in, not bolted on. People stay in charge.
AI runs through MiSAFE QHSE and is part of how the system works. It carries the administrative load. It does not run the system: your people review and approve everything it prepares, it is contained within your account, and our consultants set the limits it works within.
Less admin, without cutting corners.
AI is used where the work is largely administrative. What the law and your standards require stays exactly where it is, and a person always has the final say.
Drafts the documents
Policies, procedures, SOPs, risk assessments, training packs, checklists, print-ready signs and project reviews with lessons learned, written from your organisation profile, your own library and your records.
Reads what you upload
SDS, plant manuals, invoices, service reports, project files and emails, and auditor checklists, each titled by what it holds and turned into register entries and draft records for review.
Fills forms from a description
Describe an incident in plain words and the form picks up the people, place, injury and kind of incident.
Checks the quality
Generic controls, missing details and unfinished records are flagged so what is submitted is complete.
Prepares compliance fixes
Gaps against the Child Safe Standards, NSSAB, the NQS, ISO standards and legislation come with the fix prepared for an administrator to approve, and a new auditor checklist is mapped to what shows each question.
Watches for change
Changes to legislation and standards are picked up, with the affected documents and a suggested update.
High benchmarks on how AI is used.
AI is still relatively new. We set strict limits on what it may do in MiSAFE QHSE, and the system is built to hold it to them.
Never in control
Nothing it prepares is issued, sent or recorded as final until a person with the right role approves it.
Contained within your account
The system is built so AI works only inside your account, on your records, and never across accounts.
Not used to train
Your records are never used to train AI models.
Every touchpoint logged
Drafts, suggestions, edits and approvals are kept in the audit log.
Fixed rules the AI cannot change.
Every request to the AI, and every answer back, passes through rules written into the software. They are the same for every account, and they run in this order.
-
1
Kill switch. An administrator can pause AI for the whole account, and choose which roles may use it. Basic users are off by default.
-
2
Scope. Every request is stamped with your account. Another account’s name or record in the request is blocked, or cut out and counted.
-
3
Personal information. Emails, phone numbers, tax file and Medicare-style numbers, street addresses and dates of birth are redacted. Health, allergy, next of kin, banking and pay fields are removed. Names become Person A and Person B, and are put back in the answer.
-
4
Size and rate. Oversized requests are trimmed and bursts of requests are blocked.
-
5
Policy envelope. Every request carries train: false, retain: false and share: false with its account scope. A request without it is refused.
-
6
The answer. Every answer is checked for other accounts’ information, which is withheld, before anyone sees it.
A QHSE companion for every user.
MiCompanion talks in plain language and gets things done while the person is still talking.
- Tell it an incident has just happened and it asks the safety question first: is everyone safe, and does anyone need emergency services, with the Call 000 line. The report starts only when the person agrees.
- Report a hazard by talking it through. MiCompanion logs it, asks for photos, assigns the fix and closes it when the fix is done.
- Report something broken for the maintenance team, and see when it is fixed.
- Ask about the law and get the answer applied to your workplace.
- See what is waiting on you, and open it.
- Go straight to the right procedure, document or training, with an offer to assign it.
- Speak up about a concern, bullying or harassment through the right reporting pathway.
- It adapts to each person: brief or detailed, and in their language. Safe use terms explain what is recorded and who is told.
Taken seriously, and aligned with recognised standards.
Your records stay in your account. Access is set for each person, sign-in takes two steps, and every change is on the record.
SOC 2
Security controls aligned with the SOC 2 trust services criteria.
ACSC Essential Eight
Aligned with the Australian Cyber Security Centre Essential Eight mitigation strategies.
ISO/IEC 27001
Information security managed in line with ISO/IEC 27001.
See how the system works on your own records.
A demo built around your sites, your documents and the questions your people ask.