Agentic AI commerce, advertising, and orchestration
Key Questions
What is driving the agentic AI commerce disruption?
ChatGPT Ads reached $100M self-serve in 6 weeks with CPM halved to $25 and a shift to CPC, while only 6.6% of answers carry ads. OpenAI, Meta, Yahoo, Disney, and PubMatic are rolling out AI agents and agentic buying platforms.
How are B2B companies adopting AI agents?
Only 17% of B2B firms have deployed AI agents, with a 40% cancellation forecast due to short-term signals misleading without revenue ties. Teradata reports a 17% pilot-to-production rate and 99% scaling challenges.
What does the AI AGENT Act propose?
Senator Warner's bill mandates interoperability for AI agents to address infrastructure risks like power volatility and model access continuity.
Which brands are testing ChatGPT ads?
Bark, ButcherBox, and Vanta confirm channel pull but note missing controls and measurement gaps. B2B SaaS brands dominate while retailers largely sit out.
How is agentic buying changing metadata needs?
Agentic buying demands structured metadata, real-time traceable data, and protocols like MCP. OMS is emerging as a critical system of record for transactions.
What are the infrastructure risks for agentic commerce?
Risks include power volatility, model access continuity, and a 19-day APAC shutdown. EU rules will force Google and Apple to open platforms to third-party agents by 2027.
What revenue targets exist for OpenAI ads?
OpenAI targets $2.5B in ad revenue for 2026 but Emarketer forecasts it will fall 90% short, with the total AI ad market reaching only $5.41B by 2030.
How are platforms like Shopify and Salesforce responding?
Shopify shows AI-generated order growth and dynamic upsells with 16.2% vs 4% conversion. Salesforce launched a B2B Commerce Suite with native Buyer Agent on Agentforce.
Agentic buying, campaign construction, creative generation, analytics, and conversational commerce are moving toward full-funnel execution. New MCP/API offerings can create, optimize, and budget Google and Meta campaigns directly, but their promotional claims lack independent performance or governance evidence. Agent-ready product data, permissions, security, auditability, and reversible execution are more established priorities than proof of commercial outcomes.