AI and our Clients Data

Using AI in Business: Protecting Our Clients’ Data

Artificial intelligence (AI) is becoming an increasingly common part of everyday business software. From helping draft emails to assisting with accounting tasks, AI offers opportunities to improve efficiency, save time and streamline the way business owners work.

But as businesses adopt these tools, an important question needs to be asked: How do we take advantage of AI while protecting our clients’ confidential information?

At our business, protecting client information is a priority. Our approach is to embrace the benefits of technology while maintaining clear boundaries around the privacy of client data.

Our approach to AI

We take a cautious and considered approach to using AI.

It is becoming increasingly embedded in software we use every day, such as Xero and email. For now, that is where we are concentrating our AI Adoption. In particular, using AI features in emails to stream client workflow, so emails don't get lost in the system, questions get answered. Each day my email system can highlight clients who asked a question that hasn't been responded to - it prompts me I have emails to answer; it doesn't write the answers for me. Our job management software automatically reminds clients of queries we have that they haven't answered - which saves us living in our email constantly checking and sending reminders. Email was a fantastic invention, but turned into an unmanageble beast. I live in my email, and it's not been uncommon to lose over half the working day in it, and feel like I haven't done any work yet and its lunch time already. AI is helping speed up client responses, weed out emails that aren't even client related, let me get on with actual work. For that I thank AI, bring on more of that.

Another way AI has helped us is the increasing capabilities Xero are building into their program. For instance they now have features that flag transactions that are outside the norm. For the past 15 years we have manually scanned data when doing GST returns and annual accounts, looking for anomolies. Not 100% perfect because it is still a human casting their eye over it. Xeros AI now helps us - but it flags a lot of false positives, things that aren't even errors. If you have used it you have probably noticed it flagging duplicate bills, even though its your mobile phone bill that is the exact same amount each month and the dates are clearly different. But it is getting better. For now we use this as a support tool - run the report first to flag issues. Check those issues. Then perform our normal manual check. It isn't any slower or faster, but its a double check of your data and that is useful.

What we don't do is enter identifiable client information into separate, standalone AI tools, even with a paid subscription and learning turned off.

This distinction is an important part of our approach to managing client confidentiality, while AI is in its infancy and trust in these programs still needs to be built. For now we feel that the protections offered by our paid subscriptions to Office365, Googleworkspace, and Xero are better than those offered by ChatGPT and Claude.

The same goes for the increasing number of programs utilising AI that are available in Xero Marketplace. Apps that connect to your xero, and then run that information through embedded AI models - for now we are holding off while we get assurance that data protection continues.

Protecting client confidentiality

Client information must be handled responsibly, the use of AI doesn't change that.

Our approach has always been to protect client data and only request and hold data that we are legally obliged to do:

• Keep client data within approved systems. We do not deliberately copy client information into external, standalone AI tools. As we only keep data that we are legally obliged to do, nothing changes here.

• Understand the tools we use. AI features within our existing software are assessed in the context of their privacy, security and data-handling arrangements. We haven't automatically turned on AI features where we feel it doesn't enhance the client-Sprout relationship, or risks providing inaccurate results

• Use AI appropriately. AI can be useful for improving efficiency, but it doesn't replace professional judgement or our responsibility to clients.

• Check the output. AI-generated information can be inaccurate, incomplete or misleading. as anyone who has asked an AI model a question knows! Anything important must be reviewed before it is relied upon or shared.

• Maintain confidentiality. Our responsibility to protect client information continues regardless of the technology we use.

  
  

A practical balance between innovation and responsibility

We believe businesses should be able to benefit from technological advances without compromising the trust their clients place in them. That applies to us, your accountant, and to you the business owner.

We encourage business owners to think about how they use AI and what sensitive information they are sharing, both their own financial information and their customers.

We will continue to assess the technology we use as AI capabilities evolve, ensuring that efficiency and innovation remain balanced with confidentiality, security and our obligations to clients. After all, we were early adoptors of 100% cloud based accounting, having been a paperless office since I started in 2011, so we aren't averse to advancements. But adoption will still be cautious. Our business is built on trust, and we want to assure clients that we take their data privacy seriously, and that when they get advise/thoughts/opinions from us, it is us that gave that advise and not AI generated.

If you want to know more about how we use AI tools, especially in xero, please reach out.