AI continues to hold equal parts hype and practical applicability. Unfortunately, the noise around the former eclipses the latter. Punditry is a lot easier when it’s generic, so that’s the route often taken. Practical implementation and “how” rhetoric is much scarcer and, as it often goes, more valuable.

We got a taste of the more tactical and practical discussion around AI implementation at the recent Street Fight Live event, some of which we examined last week with Yext’s commentary. Another place where useful insights emerged was in a panel discussion from Microsoft, MediaOcean, and FAT Brands.

Starting at a high level, AI can hold much of the promise that’s often hyped, or “uncover the invisible to achieve the impossible,” says FAT Brands CIO Michael Chachula. But it’s not a silver bullet and requires some rigor. The oft-asserted notion is correct that AI is only as good as the data you put into it.

Consistency is Efficiency

For example, brands that are using AI for marketing – a common and promising use case – can streamline workflows in all the ways often discussed. However, if unrefined, these brands end up with so many hallucinations and rough results to clean up, that it ends up consuming more time than it saves.

One way around this says MediaOcean CMO Aaron Goldman is to train AI on brand guidelines and style manuals. The latter is something often used by publications and brands to govern the type of language they use… everything from fonts to whether or not the Oxford comma should be used (a heated debate!)

In the context of brand marketing, these guidelines can accomplish a surprising amount of refinement in AI models. And brands can take it even further by standardizing best practices with a set of stored prompts that work well for various functions. It’s all about not reinventing the wheel with every prompt.

Or as Chachula says “Consistency is efficiency when it comes to AI.”

Moving Target

Lastly, it should be noted that all the above is a moving target. AI underlying technology and capabilities are evolving so quickly – as they’re designed to do – that the role of AI itself will dynamically adjust. The lesson for brands and enterprises is to anticipate that future role and skate to where the puck is going.

That playbook will be all about the lessons espoused above: AI’s value will lie not in the underlying models – which will increasingly commoditize – but in how you use them and the data you input. This is perhaps best summed up by Microsoft’s Joe Veverka, quoting his boss Satya Nadella:

“As AI becomes more capable and agentic, models themselves become more of a commodity, and all value gets created by how you steer, ground, and finetune these models with your business data and
workflow—and how they compose with the UI layer of human to AI to human interaction.”