Skip to main content

Heutrix AI Guardrails for responsible day-to-day use.

Commercially available AI tools can support internal admin, drafting and workflow guidance. They can also create privacy, quality, security and accountability risks when staff use them without approved boundaries.

Heutrix helps organisations turn general AI concerns into clear rules for AI use, information boundaries, human review responsibilities and practical staff guidance.

What Heutrix AI Guardrails helps with

The goal is a usable internal position: what staff may do, what they must not do, what requires approval and who remains accountable for the final output.

  • Which AI uses your organisation may allow, restrict or prohibit
  • Information that must not be entered into public AI tools
  • Tool-selection and access considerations
  • Who checks the results and when to seek help
  • Responsibility for approving AI-assisted work
  • How staff should check and document outputs
  • Rules to confirm before wider rollout

AI uses to consider

Depending on the information, tool and review process, AI may assist with:

  • Drafting internal admin templates from non-sensitive prompts
  • Generating first drafts of process checklists
  • Summarising non-sensitive internal notes
  • Preparing first drafts of internal workflow guidance
  • Structuring internal administrative reporting templates for human review
  • Supporting approved internal knowledge searches
  • Creating plain-English explanations for internal training
  • Organising non-sensitive tasks or next-step suggestions

Suitability must be assessed for the specific purpose, information, tool, user and review process.

AI uses outside this service

Qualified review does not bring clinical decision-support or client-facing clinical or support advice into this non-clinical service.

  • Replace clinical judgement
  • Replace legal advice
  • Replace organisational or compliance accountability
  • Make final clinical, safety, staffing, risk or compliance decisions
  • Make incident or complaint outcome decisions
  • Enter personal or sensitive information into publicly available AI tools
  • Process sensitive information without a documented and approved process
  • Generate client-facing clinical or support advice
  • Remove human responsibility for checking outputs

Example AI use positions

These examples are starting positions only. Your organisation approves its own tools and AI uses.

AI use type

Drafting internal admin notes from non-sensitive prompts

Position

Assess tool, information and approval first

AI use type

Summarising internal meeting notes

Position

Assess the content and tool before use

AI use type

Creating a first draft of a process checklist

Position

Assess tool, information and approval first

AI use type

Writing client-facing clinical or support advice

Position

Outside this non-clinical engagement

AI use type

Entering identifiable information into publicly available AI tools

Position

Do not use

AI use type

Using AI to make final clinical, staffing, compliance or safety decisions

Position

Not suitable

Privacy and sensitive information

Personal information entered into an AI product remains subject to the organisation's privacy and data handling responsibilities. Publicly available AI tools should not receive personal or sensitive information.

Any proposed AI use involving personal information needs a documented, organisation-approved process covering necessity, tool suitability, access, storage, retention, permissions and accountable human review.

What is included

  • Current and intended AI use review
  • A record of the agreed AI uses to review
  • Proposed rules for allowed, restricted and prohibited uses, for your organisation to approve
  • What information staff can and cannot share with AI
  • Who checks and approves AI-assisted work and when to seek help
  • Staff AI use checklist
  • Prompt guidance for approved non-sensitive use
  • Output review checklist
  • Tool and rollout recommendations

Heutrix AI Guardrails provides guidance for agreed admin uses. Your organisation approves and maintains the rules. Legal, privacy, clinical and regulatory advice are outside this service.

Free tools and workbooks

A practical place to start.

Choose how you want to work: use the online assessment for a result you can discuss with your team, or download the Excel version to work through the problem at your own pace.

Results and Excel downloads are free. No email required. Keep information general and non-sensitive. These resources are a starting point for discussion, not tailored advice or approval to implement a change.

Free online tool

Clarify the boundaries for AI

AI Guardrails Check

Review one planned use of AI in admin work, including information boundaries, approval and checks by your team.

See which controls need attention before your team proceeds.

Or work in Excel

Record planned AI uses, decisions and conditions, with guidance included.

Download the Excel Register

Set clear AI rules before staff start relying on AI at work.