[Column] Keith Kakadia: AI chats drive social commerce

As Meta expands its business AI plans, brands face a practical test: can an automated conversation help someone buy without giving them the wrong answer?

On September 28, Meta announced the Meta Enterprise Platform and named former MongoDB CEO Chirantan "CJ" Desai its chief enterprise platform officer. The company said its initial focus includes bringing Meta Business Agent, its Muse agent, and other AI tools to businesses. It has not been announced that the new platform will change how Instagram or Facebook ranks brands or products.

The near-term commerce story is Meta Business Agent, which the company introduced in June. Meta says the agent can answer business-specific questions, recommend products from a catalog, book appointments, and pass a conversation to a person. It says more than one million businesses already use a Business Agent on WhatsApp and Messenger.

Imagine someone sees a jacket in a Reel and messages the brand to ask whether it runs small. If the agent answers quickly but gets the sizing wrong, the brand has created a return, not a better shopping experience. That's the sort of conversation I'd test first.

The stakes extend beyond response speed. In a September IAB study of 2,200 people across five countries who had recently used AI for shopping research, 65% said credible creator reviews made them more confident in AI recommendations. The finding suggests that useful recommendations need evidence that a shopper can assess, such as a reviewer's experience with the product. It does not establish that Meta's agents use creator reviews in their answers.

For brands considering an AI agent in social messaging, Kakadia recommends starting with three checks:

Ask the questions customers actually send. Test fit, availability, shipping costs, and return conditions, including cases where the right answer is "I don't know."

Check the handoff. Make sure a customer can reach a person when an answer depends on an exception or the conversation turns into a complaint.

Review the outcome. Look beyond the number of chats resolved. Check whether customers received accurate information and whether the interaction led to a sale, a return or another support request.

Brands already have a record of what confuses people in the comments and inbox. Use those questions as the test set. A polished demo won't tell you how the agent handles an out-of-stock item or a customer who needs an answer before tonight.

That scrutiny matters as marketers invest more in AI. An IAB survey found that 83% of advertising executives said their companies had deployed AI in the creative process, while only 45% of surveyed Gen Z and Millennial consumers felt positive about AI-generated ads. The study concerns advertising rather than customer service, but it is a reminder to measure the customer's experience alongside the business's efficiency.

Meta's announcement gives businesses another reason to examine where AI fits in their operations. In social commerce the first standard should be straightforward: would the customer trust the answer enough to act on it?

Keith Kakadi is a social media expert and founder of Sociallyin

Follow Marketing Report on LinkedIn