As part of its AI blitz over the past two weeks, Meta has teased new chatbot features with particular implications for brands. The key development is that marketers will be able to train chatbots tailored to their individual brand personalities to use as everything from customer service agents to quirky entertainers across Meta’s messaging apps. 

The feature is currently being tested by a small group of unidentified businesses, although no release horizon has been given. Meta did not return a request for comment for this story.

Virtual assistants, of course, have long been used in e-commerce, but the difference with these new bots is the sheer power of the AI supporting them. No longer annoyingly narrow in scope, chatbots are now capable of quickly generating responses for a seemingly infinite variety of questions, and in a humanlike manner. 

With the added ability of Meta’s bots to calibrate this power to a brand’s specific voice and set of needs, marketers are intrigued.

E.l.f. Cosmetics is “absolutely” interested in testing a branded bot, and has requested to be part of the pilot, said Ekta Chopra, e.l.f.’s chief digital officer. Top of mind is Meta’s proposition that its AI bots can be tailored to a brand’s tone, which has so far been a shortcoming of AI chatbots, said Chopra.

“We don’t want it to sound like a robot,” she said.

An opportunity may exist for brands that already have a mascot or central character, such as Wendy’s or Progressive. These marketers could aim to channel their mascot’s personality through the bot, giving consumers a reason to engage, said Damian Rollison, director of market insights for martech company SOCi. 

“I could imagine a Wendy’s bot that roasts you—that would be entertaining,” he said, alluding to the chain’s popular social media antics.

Watch Wendy’s best TikTok zingers for ‘National Roast Day’

Another draw of Meta’s bots is that they would be immediately available in some of the most widely used consumer platforms in the world. Chopra said that e.l.f. would be particularly interested in offering a bot on WhatsApp in order to drive the brand’s international reach. The messaging platform owned by Meta now caters to over half of the world’s internet users, according to CNBC.

PacSun is open to creating an AI bot on Instagram, where it already has a strong presence with customers, said Shirley Gao, the retailer’s chief information officer. Doing so could protect against the kind of customer attrition that occurs when buyers have to leave an app and open a new one in order to continue their purchasing journeys. Meta has not yet announced plans for bringing its AI capabilities to Instagram, although an expansion is ultimately expected.

A pivotal question will be whether Meta can convince brands to not only try out the bots but also to continue using them. Some marketers may remember Meta, back when it was Facebook, launching a bot platform on Messenger in 2016. These comparatively simple AIs were meant for handling customer service, and while initially garnering attention from brands, they fell out of favor.

AI companies such as Haptik have picked up where Meta left off, helping brands develop more effective bots on platforms including WhatsApp. But Meta’s technological resources afford it a unique opportunity to attract marketers back to its chatbot solutions. Haptik did not return a request for comment.

AIs for entertaining

As part of its effort to make chatbots more appealing, Meta demonstrated how celebrities could be incorporated by endowing their face to an AI character. At its annual Connect conference last month, Meta unveiled a series of these star-powered AIs, including NFL legend Tom Brady as an ardent sports bot named Bru, with whom users are encouraged to debate.

Brands have a direct path toward leveraging stars as the faces of their bots given the popularity of celebrity-brand partnerships. Gao said that PacSun would be intrigued by such a possibility because it could strengthen a connection with Gen Z customers.

Top 5 celebrity-brand collabs to know right now

Celebrity-embodied bots are not yet being tested by brands, nor has Meta announced functionality just yet for allowing marketers to bring in their own celebrity partners. But these AI bots hit on an important point about Meta’s new offerings—namely, that they’re supposed to be fun for users.

“This isn’t just going to be about answering queries,” Meta CEO Mark Zuckerberg said at Connect. “This is about entertainment and about helping you do things to connect with the people around you.”

The draw for content creators is obvious: Fans will be able to chat with virtual versions of their favorite personalities. Brands, however, are not as beloved, which makes the value proposition less clear.

An ad tech expert, speaking on the condition of anonymity, expressed skepticism that entertainment would be a sufficient motive for consumers to engage with businesses in this way. 

“Brands that use direct-to-consumer commerce should look at this first, because otherwise the consumer is just talking to a branded bot,” they told Ad Age. 

SOCi’s Rollison is unconvinced for a similar reason. Marketers, he said, will need to foreground the practical use cases—such as customer service—in order to drive any engagement. 

Gao, on the other hand, sees providing a fun experience as enough reason to test out such a feature. She pointed to previous, pop culture-inspired activations done by PacSun, such as a Roblox world and an NFT collection, as precedents for this motive. 

“It makes the customer feel that we’re connected,” she said.

More to worry about

Even if a brand is able to team up with a beloved celebrity and create the most entertaining AI in the world, worries abound.

Teaming with celebrity-powered bots could expose marketers to controversial territory, such as the dilemma around a person’s rights when issuing their digital likeness to AI, Rollison said. 

This issue came to a head during the recently resolved Writers Guild of America (WGA) strike and still ongoing Screen Actors Guild-American Federation of Television and Radio Artists (SAG-AFTRA) strike. In the latter movement, actors have fought to prevent their likeness from being exploited via unconsented and uncompensated digital replication. 

Should Meta allow brands to bring in their own celebrity partners, the same apprehension will likely apply. The stars who signed on to Meta’s initial batch of AIs were given a pretty penny for doing so, in some cases netting $5 million for roughly six hours of work, according to The Information. 

But these stars lent only their faces. Remuneration could become exorbitant—perhaps even priceless—as Meta expands the degree of embodiment that is possible through these AIs, beginning with voice enablement sometime next year.

With this uncertainty in mind, e.l.f. is hesitant to test any AIs tied to real people until the debate over likeness is resolved, Chopra said.

AI hallucination and a lack of regulation will continue to cause other brand safety issues. For example, some of the celebrities embodying Meta’s bots are dissing their real-life brand partners, as reported by Fortune. 

Even Nick Clegg, Meta’s president of global affairs, is exercising caution over the bots’ tendency to spew unwelcome responses. The company has been trying to protect against these outcomes through thousands of hours of stress testing, according to Bloomberg News.

“It’s not super smart to start allocating your budget to this today,” Rollison said.