Sharon Gee Urges Retailers to Prepare for Agentic Commerce

Sharon Gee spoke at Commerce Live about the latest operational and technical developments in Agentic Commerce and what brands and retailers need to do now to be ready for a UK rollout. She framed the problem as an integration and trust engineering challenge: autonomous agents will be able to discover, negotiate, and transact, which requires merchants to expose dependable, secure transactional endpoints and clear policy controls.

Agentic Commerce is the operationalization of autonomous consumer agents interacting with commerce systems, requiring durable agent orchestration, intent schema, and transactional APIs. Engineers must support real-time inventory and pricing, deterministic payment tokenization, and atomic order flows to avoid double-sells and unintended charges. Practically, teams should:

• •Update APIs to be idempotent, versioned, and support agent-specific scopes
• •Normalize catalogs and expose real-time availability and pricing feeds
• •Instrument rich telemetry for intent, decision paths, and outcome auditing
• •Harden payment and fraud systems with agent-aware rules and tokenized flows

Why it matters This is a structural shift from conversational commerce to autonomous execution. Models and agents will depend on high-quality retrieval from product catalogs, embedding-based personalization, and low-latency transactional primitives. Data scientists must adapt ranking and recommendation pipelines to operate under agent-driven queries that optimize for multi-step objectives, not single-click conversion metrics. Operations and legal teams must also define explicit consent models and approval gates so agents can act within merchant and regulatory boundaries.

Practical risks and mitigations Merchant risk expands beyond UX to include liability for agent-initiated purchases, increased fraud surface, and supply-chain inconsistencies. Mitigations include staged rollout using shadow mode simulations, synthetic-agent testing, stricter authentication for high-value intents, and verbose audit logs to support dispute resolution. Monitoring should track conversion integrity, transaction success rate, and a rising incidence of exception flows.

Track pilots that open payment and fulfillment APIs to agents, evolving regulatory guidance in the UK on agent consent and liability, and standards for agent metadata and provenance. For practitioners, prioritize API hardening, telemetry, and simulated agent testing in the next 3-6 months to remain operationally ready.