Enterprise commerce and AI orchestration
commercetools
commercetools builds enterprise commerce software for businesses selling across channels, regions and brands. Its Sphere platform connects commerce APIs with governance for AI agents, letting enterprises control access through permissions, policies and limits. Standalone Core Commerce and Product Catalog provide cart, checkout, order and customer management, alongside product modeling, pricing, inventory and search. Enterprises can adopt these capabilities individually while retaining the rest of their commerce stack.
Founded by Dirk Hoerig and Denis Werner, commercetools developed a headless, API-first architecture that separates commerce capabilities from customer-facing experiences. Doug McNary is CEO, while Hoerig serves as chief innovation officer. Its MosAIc orchestration layer extends this approach into business workflows: a plain-language request becomes a plan coordinated through agents across commercetools and third-party systems. Generative interfaces assemble the relevant data and controls for each task—for example, displaying products at risk of selling out with stock-cover estimates and reorder controls.
MosAIc’s campaign optimizer illustrates how those workflows operate: users review and approve a campaign’s budget, markets and products, then the agent updates personalized offers daily using incoming orders. It stays within configured limits and pauses when the budget is reached. In 2026, commercetools said MosAIc was live with a select customer group. That year, the company reported that its broader commerce platform served more than 600 brands, retailers and enterprise companies, processing more than $110 billion in annualized gross merchandise value.
Explore the recordings
This commercetools archive contains one recording, with two useful paths through it: intent-driven UI architecture and the design contracts that make generated interfaces predictable. The guide describes claims made in the recording; it does not establish commercetools’ current implementation or product status.
From inconsistent screens to constrained UI generation
Start with The End of the Static Screen: Architecting Intent-Driven UX — Gus Iwanaga for an architectural case study. Iwanaga describes an early approach that produced inconsistent interfaces for repeated requests for the same sales report, which he judged unsuitable for production. He then explains a revised pipeline: classify intent, invoke tools, retrieve data, select eligible components, and generate a schema-constrained UI description rendered with native React components. His comparison of fixed components, sandboxed HTML, and declarative UI offers a useful way to examine the tradeoff between predictability and model autonomy.
The component catalog as a design contract
Return to Iwanaga’s talk through its discussion of information architecture and component schemas. He describes teaching a UX agent layouts and templates based on atomic design, mapping components through sub-slots and slots. For designers and product teams, the useful thread is his account of work shifting toward schemas, rules, synthetic data, component mappings, and interaction patterns. He presents coherent layouts and careful catalog-property curation as continuing challenges at the time of the recording.
1 talk
Newest first
1 speaker at AIE
Affiliations reflect their AIE appearances, not necessarily current employment.