Floward deploys agentic AI to absorb 13x peak gifting demand

Floward, the online flower and gifting retailer operating across the Middle East and the UK, has deployed a multi-agent AI customer service stack built on Infobip's AgentOS platform, enabling the business to absorb conversation volumes up to 13 times higher than baseline on peak gifting days without a proportional increase in headcount.

The deployment is a live stress-test of the "agentic workforce" thesis that has attracted significant venture and corporate investment across retail, logistics and financial services over the past 18 months. For Floward, the stakes are concrete: as a same-day delivery business, its revenue is structurally concentrated around a handful of high-intensity occasions including Valentine's Day, Mother's Day and Ramadan. A single day of degraded service during those windows carries disproportionate brand and revenue cost.

From rule-based bots to orchestrated agents

Working with Infobip's AI consultants, Floward redesigned its customer journey architecture over fewer than two months. The new system replaces traditional rule-based chatbots with a multi-agent setup in which an orchestration layer routes each incoming conversation to a specialist AI agent: one handles address collection, another manages order amendments, a third serves frequently asked questions. When a query exceeds the AI's scope, it escalates to a live agent within the same session.

A notable structural choice was the use of WhatsApp as the primary channel. Gift recipients are guided through address confirmation directly inside a WhatsApp conversation, while the AI simultaneously handles related queries in the same thread. This keeps the customer inside a single interface rather than bouncing them across email, a web portal and a support ticket system. On Valentine's Day 2026, the platform handled 54,000 conversations in a single day, sustaining a one-minute average response time and a 95% SLA compliance rate. Floward's Customer Care Senior Manager, Lujain Mallosh, reports a 15% reduction in customer service costs and a 12 percentage-point improvement in customer satisfaction scores versus the previous peak period.

The macro read-across: agentic AI as a retail infrastructure layer

The Floward-Infobip case is worth reading beyond the gifting vertical. What the deployment demonstrates is a pattern now emerging across occasion-driven and seasonally concentrated retail categories: AI agents functioning less as a cost-cutting overlay and more as a scalable infrastructure layer that decouples revenue capacity from headcount growth.

For the cross-sector investor, the signal is in the build time and the channel choice. A production-grade, multi-agent customer service system deployed in under two months using an existing communications platform represents a meaningful compression of the enterprise AI adoption cycle. The WhatsApp channel selection also reflects a broader structural reality in the GCC and South Asia consumer markets, where WhatsApp penetration functions as a de facto business communication standard rather than an optional add-on.

Infobip is extending its AgentOS logic into revenue-generating territory at Floward. The next phase includes an e-invitations feature in which approval workflows, recipient notifications and gift-prompting sequences are all orchestrated through the same platform. That expansion moves AgentOS from a cost-centre tool into a growth-surface tool, a transition that carries implications for how communications platforms are valued relative to pure CRM or contact-centre software vendors.

The broader capital landscape reflects this repositioning. Agentic customer experience platforms have attracted investment from both enterprise software strategists and retail-vertical funds over the past year, with deal flow accelerating in the GCC as regional businesses invest in digital infrastructure capable of handling hypergrowth demand curves. The Floward case adds a documented, metrics-backed data point to that investment thesis, at a moment when many comparable deployments remain in pilot or proof-of-concept phase.