Bluebell Admin Services Ltd
AI Usage & Transparency Policy
Trading as Bluebell Bookkeeping & Admin
Last reviewed: September 2026
1. Why we have this policy
Bluebell uses artificial intelligence (“AI”) as a tool to support some of the work we do.
AI can be extremely useful. It can help us analyse information, spot patterns, summarise large amounts of information and work more efficiently. But it is a tool – it does not replace professional judgement, experience or responsibility.
AI assists. We review, challenge and remain accountable.
This policy explains how we use AI, the safeguards we put around its use and what that means for our clients and other people whose information we handle.
2. How might we use AI?
We may use approved AI tools to help with tasks such as:
- analysing financial or business information;
- identifying patterns, anomalies or inconsistencies;
- assisting with calculations and working papers;
- summarising documents or information;
- research;
- drafting or improving documents and communications;
- organising or explaining information;
- meeting transcription or summarisation; and
- administrative and business-support tasks.
We may also use purpose-built AI assistants configured to support particular areas of our work. For example, we use an HMRC-focused assistant to help with tax and compliance research and to work with relevant HMRC guidance and information.
These tools remain assistive. Their output is reviewed and, where appropriate, checked against current HMRC guidance, legislation or other authoritative sources before we rely on it.
AI is there to assist our work, not to replace the person responsible for doing it.
We may not use AI for every client, every service or every task.
3. Human review and professional judgement
We don’t assume something is correct simply because an AI system produced it.
Where AI is used as part of our professional work, its output is reviewed as appropriate for the task. This may include checking it against source records, calculations, legislation, professional guidance or our own knowledge and experience.
If an AI output doesn’t look right, we question it.
Where something requires professional judgement, that judgement remains with us.
We do not use AI to make solely automated decisions about people that have legal or similarly significant effects.
4. Personal and client information
AI tools may sometimes process personal, financial or business information where this is reasonably necessary for the task being carried out.
We don’t pretend otherwise.
Where we use information with AI, we aim to use only what is reasonably necessary and to minimise, anonymise or remove identifying information where doing so is practical and doesn’t undermine the task.
We do not deliberately provide AI tools with unnecessary personal information, passwords, login details, authentication credentials or other information that they do not need.
Our use of personal information with AI is also covered by our Privacy & Data Protection Notice.
5. Confidential information
Client confidentiality still applies when AI is involved.
We consider the nature of the information, why it is needed and whether using an AI tool is appropriate before using confidential information with AI.
We use approved tools and take reasonable steps to understand how those providers handle and protect information.
AI doesn’t create a free pass to share information that we wouldn’t otherwise be entitled to use or disclose.
6. Accuracy
AI can make mistakes.
It can misunderstand information, miss context or produce an answer that sounds convincing but is wrong.
That’s why AI output is not automatically treated as fact.
The amount of checking we carry out will depend on what the AI has been used for and the potential consequences if the output is wrong.
For example, an idea for wording in an email does not carry the same risk as analysis contributing to a VAT calculation.
The greater the potential impact, the greater the level of human checking we expect.
7. Fairness and bias
AI systems can sometimes produce biased, incomplete or misleading results because of the information they were trained on or the way they operate.
We remain alert to this and do not knowingly use AI in a way that unfairly discriminates against an individual or group.
Where an AI output could affect a person, we consider whether the result is reasonable and appropriate rather than simply accepting it.
8. Security and approved AI tools
We only use AI tools that we consider appropriate for the work being carried out.
When deciding whether a tool is suitable, we consider matters such as its purpose, security, privacy arrangements and how information provided to it may be used.
Where appropriate, access to AI tools is restricted to authorised people and appropriate account and security controls are used.
9. International processing
Some AI providers may process information outside the UK or use international infrastructure.
Where personal information is involved, our approach to international processing is the same as for our other technology providers: we take appropriate steps to ensure information remains protected in accordance with UK data protection law.
More information about international processing is available in our Privacy & Data Protection Notice.
10. Intellectual property and copyright
AI-generated material can raise questions about copyright, ownership and the use of third-party material.
We don’t assume that something produced by AI is automatically ours to use however we like.
Where relevant, we consider intellectual-property rights and take reasonable steps to avoid knowingly using AI-generated content in a way that infringes another person’s rights.
Likewise, we don’t deliberately provide copyrighted or confidential material to an AI system where we don’t have an appropriate reason or right to do so.
11. Transparency
We believe clients should be able to understand how AI fits into the way Bluebell works.
That doesn’t mean we need to announce every time AI has helped reword a sentence, summarise some notes or perform another routine supporting task.
It does mean we won’t deliberately hide material use of AI where knowing about that use would reasonably matter.
If you have questions about how AI has been used in connection with your information or the services we provide, you can ask us.
12. Responsibility
Using AI does not transfer responsibility for our work to the AI provider.
Bluebell remains responsible for the services we provide and for the decisions and professional judgements we make.
That remains true whether a piece of work involved a calculator, spreadsheet, accounting system, AI tool or any other technology.
AI assists. We review, challenge and remain accountable.
13. Questions or concerns
If you have a question or concern about our use of AI, please contact:
Bluebell Admin Services Ltd t/a Bluebell Bookkeeping & Admin
Hitherwood
The Street
Rotherwick
Hampshire
RG27 9BL
Email: info@bluebelladmin.co.uk
Telephone: 01256 268998
If your concern relates specifically to how we have handled your personal information, our Privacy & Data Protection Notice explains your data-protection rights and how to make a complaint.
14. Keeping this policy up to date
AI technology, regulation and guidance are developing quickly.
We therefore review this policy regularly and may update it when our use of AI, the technology we use or the rules applying to it change.
Last reviewed: September 2026
Next review: September 2027, or sooner if something significant changes.