The Drum Awards Festival - Extended Deadline

-d -h -min -sec

Artificial Intelligence Ecommerce Technology

What brands need to watch out for when launching AI-powered assistants

By Arif Miah, Creative strategy director

mud orange


The Drum Network article

This content is produced by The Drum Network, a paid-for membership club for CEOs and their agencies who want to share their expertise and grow their business.

Find out more

May 24, 2023 | 7 min read

With brands jumping on the AI assistant bandwagon, Arif Miah of Mud Orange warns of the implications when done without consideration for the brand’s guidelines or customer needs.

Lego figure holding a wrench

Brands that want to have a stab at AI features should proceed with caution / Marcus Ganahl

The last couple of months has seen a wave of brands milking ChatGPT to create their own artificial intelligence (AI)-powered assistants. As you’d expect, these AI assistants are preloaded with prompts (rules on what to talk about and how to talk about it) for customers to interact with and use.

But the rush to ‘innovate’ has exposed brands to some pretty hefty risks which, if not maneuvered carefully, could lead to a disjointed and incoherent experience that distracts from the core brand offering. On that note, here are some key watch-outs before brands launch their own AI tool.

Powered by AI

Explore frequently asked questions

Support the offering, don’t distract from it

When launching an AI-powered feature, it needs to be strategically integrated into the overall brand ecosystem and support their customers’ real needs. There have been loads of newly launched half-baked AI assistants that just feel gimmicky and unnecessary, which can make the customer experience clunky and confusing as it overshadows the core product.

The role of AI-powered features should be to complement and enhance the existing brand offering. A great example is the upcoming launch of Zalando’s fashion assistant which allows customers to ask questions to find the right fashion pieces through their own language, helping people navigate Zalando’s database of items in a conversational way. Zalando’s extensive catalog of items has long been a pain point for customers, making the feature strategically in line with its customers’ needs.

Screengrab of Zalando's AI feature

Align with the brand

As brands rapidly dispatch AI chat features, they often fail to account for language, personality, and tone of voice guidelines, which undermines the huge amounts of work across the brand world. To drive consistent brand associations and brand recognition, brand essence needs to be upheld across all touchpoints.

Memrise, a language learning platform, has done a great job in this when building Membot, an AI-powered language partner that helps people practice their second language stills. The way Membot interacts with users is fun, practical and stimulating, making practicing a new language much more approachable – just like how the brand positions itself.

At Mud Orange, we’re working with a tech brand, building their first ever ‘AI Usage Guidelines’. The truth is, many people within brand teams are already using AI tools for a range of tasks, from social media updates to emails for new partners. So just as we would create ‘Brand Guidelines’ or ‘Social Style Guides’, we’re helping our clients get in shape to make the most of AI.

Don’t AI for AI’s sake

With all the AI hype in the business realm, applying it in the public realm needs more branding attention. The prominence of Open AI’s Chat GPT has given false confidence to others that it’s acceptable to use ambiguously technical names for projects without giving due attention to how it’s named.

Good examples are GrammerlyGo, Spotify DJ, Ask Klarna, and GitHub Copilot. Having an ownable name allows a brand to distinguish itself from alternatives and helps customers navigate across all these new AI launches. Notion AI and Snap’s ‘My AI’ is just lazy and the user feedback for both echoes users’ confusion.

Screengrab of Spotify's 'Meet Your DJ' AI feature

And finally, test, test, test

With AI having a mind of its own, brands need to make sure they stress test and gain customer feedback thoroughly before launching. That doesn’t mean expensive and time-consuming studies, but some of those released have felt premature. ‘My AI’ from Snapchat, for example, has raised alarm bells around privacy concerns and strange messaging patterns.

Away from chat assistants, Levi’s use of AI models to increase diversity and inclusion has been criticized for being tone-deaf and a poor gesture as it fails to understand the impact true diversity brings. The AI rush to market can cause damage to brand reputation and undo years of strong brand building.

With proper testing and consumer feedback, key decision-makers can better understand where and how AI will resonate, helping them make suitable solutions that add real value to their customers.

Artificial Intelligence Ecommerce Technology

Content by The Drum Network member:

mud orange

mud orange is an independent creative agency producing award-winning work, that works with niche audiences. Based in London, they’ve built a global name for producing...

Find out more

More from Artificial Intelligence

View all


Industry insights

View all
Add your own content +