AI guardrails: Where AI should not be used – and why that matters just as much
Knowing where AI should not be used is just as important as knowing where it should. Here is how we set the guardrails together with our clients.
Most of what is written about AI is about what you can use it for. But when we work with businesses, it is at least as important to agree on where AI should not be used. That is why “AI guardrails” is a fixed part of the report you receive after a Shadow Walk.
Why guardrails?
Without clear guardrails, one of two things usually happens. Either AI is used everywhere, including where it creates risk. Or nobody dares to use it, because nobody knows what is allowed. Both are costly. Clear guardrails actually make it easier to get started, because everyone knows where the limits are.
What the guardrails contain
The guardrails are concrete and written for your business. They typically cover three things:
- Where AI should not be used. This could be decisions that have a big impact on individuals, or tasks where mistakes are expensive and hard to spot.
- Where a person must approve. AI can easily draft a quote, a report or a reply to a customer. But a person must review it before it is sent.
- Principles for data. Which data may AI use, where may it be stored, and who has access? For many businesses it is essential that sensitive data stays on their own systems.
Guardrails that fit the working day
We do not write the guardrails from a standard template. We start from what we saw during the Shadow Walk: which tasks are done, which data is used, and where mistakes have consequences. That makes the guardrails easy to understand and easy to follow, because they are about your work and not about AI in general.
The guardrails also tie in with the rules that apply to you, such as GDPR and the EU AI Act. We are not lawyers, and the guardrails do not replace legal advice. But they give you a practical starting point that your lawyers or advisers can build on.
Good guardrails do not slow down the use of AI. They make it safe to use.
Getting started
AI guardrails is the last of the five documents in the report after a Shadow Walk. Together with the Friction Map, the prioritised opportunities, the recommendation for what to build first and the plan, it gives you a complete basis for deciding how you want to use AI. Get in touch if you would like to hear more.
Want to see what it could look like for you?
We are happy to have a no-obligation talk about your day-to-day work and where AI and software can make a difference.
