I’m not a technologist by trade. I’m a strategist, and I’ve always – probably in blinkered arrogance – considered that the most valuable work we do at Landor sits at the intersection of strategy and creative.

But this past year, I’ve had a Damascene conversion.

As I’ve worked with WPP’s AI specialists, Satalia – exploring what’s possible through new intelligent marketing models, as well as what our clients want from them – I’ve concluded that brand guidelines are about to have their moment.

In the AI arms race, data is our most precious resource. When we picture data, we often think of numbers. But data is any information that’s used to train an AI model. In the case of AI models that enforce brand compliance and generate on-brand assets at speed and scale, the data they feed on are rules.

Rules will be the bedrock of all brand-based artificial intelligence. Because rules will help compliance models identify misuse and ensure on-brand execution. And then it’s only a small leap to generative intelligence.

Bingo! I thought, reflecting on the fact that we – at Landor – are sitting on millions of pages of brand guidelines. Surely, an overlooked treasure trove, ready for mining!

But as we started to test, I quickly came to realise that our human-made guidelines were not quite as AI-ready as I had hoped.

That’s not to say that the guidelines weren’t good. For humans, many were optimal. But not for AI.

They were too heterogeneous, too creative, too artisanal. Someone has assumed too much knowledge here and left too much room for interpretation and creativity there.

In an AI age guidelines require an overhaul.

So, what now?

Here are three very practical learnings that I’ve identified since starting work on AI compliance, to get your guidelines AI-ready.

1. Assumed knowledge is our enemy

What do I mean by “assumed knowledge”?

I mean that humans, especially experts, tend to assume a common level of knowledge. For designers and brand experts, these are unwritten principles that they absorbed at design school, or over the years at work in an agency.

But, shocking as this may sound, AI never went to design school. Nor has AI spent years working in a branding agency. AI, as naturally well-suited as it might be for brand compliance, is rather a novice.

We must start from the point where nothing is obvious, and everything is explainable.

Whatever we assume is known, we must forget, and walk through even the most basic elements in forensic detail.

That means that AI-ready guidelines might seem overly simplistic, or overly detailed to the human eye. But that’s what it takes.

2. Get your taxonomy in check. And stick to it

As the science of organising information, taxonomy is what lets us categorise large amounts of data so it’s easier for AI to understand and process.

The simpler and more standardised your taxonomy, the easier you make it for your AI model to learn and deliver high-quality outputs.

The problem is that guidelines – having largely been created by individuals, over many years – are rather an artisanal venture. Even when done by big agencies, with standard processes. Nothing is made completely to scale. There is always a bespoke element.

We are now working on a common taxonomy, to ensure there is a standard approach to categorising rules within guidelines.

This will make it much easier for AI models to learn in the future. But success will rely on strict adherence.

So, while creativity is essential in an AI future – guidelines might not be the place to exercise it.

3. No AI-build before guidelines rework

Now, I fully understand the excitement and temptation to want to build something. To see an AI model deliver results for you, so you can show it off to your teams, to your boss.

But my strong advice after a year of exploring this space, would be not to build anything at all until you have your data in order.

That means, if you want to build brand compliance, or on-brand generative solutions, you need to spend a good chunk of time, effort and resource rethinking your guidelines.

The worst thing you could do is start building without comprehensively reviewing and optimising the existing rules – overriding assumed knowledge and getting your taxonomy in check.

Because in an AI world, you get what you give.

A little bit of preparedness will go a long way to unlocking huge global economies of scale.

So, I leave you with the thought that – if guidelines are the new gold – find a way to invest in them to set your brand up for success in the future.

Rules need to be rewritten for an AI world. Guidelines need to be made AI-ready.