Meta launches Business Agent for in-app shopping and support

Meta has launched Business Agent, an AI service that automates shopping and customer support inside its messaging apps. The agent is available in Instagram and Messenger now and will arrive on WhatsApp in the near future. Retailers can use it to complete purchases and resolve routine support requests without human intervention.

The software embeds a persistent agent in conversations. When a shopper asks about a product, the agent can answer sizing questions, suggest alternatives using the retailer’s product data, guide the buyer through checkout inside the app and process payment without redirecting to external portals. Meta says the design reduces cart abandonment linked to leaving the host application.

Business Agent integrates directly with merchants’ product catalogs and business information so recommendations reflect current inventory and pricing. The underlying models learn from ongoing interactions and can improve performance over time, reducing the need for manual reprogramming. Retailers with seasonal catalogs and volatile demand can update offerings more quickly through automated syncing.

Meta positions the agent to handle high volumes of repetitive tier-one tickets around the clock, allowing human staff to focus on complex account issues and retention work. Native integration with a user’s social graph and interaction history enables more specific recommendations and routing than many third-party tools can provide.

Deployment requires operational and security preparations. Companies must supply clean, machine-readable product and support documentation, because poorly structured input data can produce incorrect or damaging customer interactions. Engineering teams need to build clear escalation paths and handover protocols so complex or exceptional cases move promptly to human agents. Quality assurance teams should test escalation triggers and run large numbers of simulated conversations to find edge cases and prevent customers from becoming trapped in automated loops.

Authentication and identity verification are required for returns, refunds and order-status queries. Retailers must integrate Business Agent workflows with existing single sign-on systems and enterprise authentication methods to ensure secure access before processing sensitive transactions. Firms also need to define operational limits for the agent to prevent unauthorized internal actions.

Choosing Business Agent involves weighing platform dependency against building or maintaining an independent system. Using Meta’s native agent reduces initial development costs and offers immediate access to the platform’s user base and processing infrastructure, which may suit small and medium businesses. Proprietary engineering stacks demand higher ongoing maintenance but provide more control over model selection, data residency and long-term portability. Many organisations are likely to use hybrid setups in which Meta’s agent handles discovery and routine routing while high-value financial transactions and complex account resolutions are routed to internal, secured systems.

Large enterprises will need to assess how the managed service aligns with existing CRM systems and compliance rules, while smaller retailers can use the lower technical barriers to deploy conversational commerce faster.