Why Sitecore Picked the AI SDK for Its Brand Assistant
Sitecore, a digital experience platform, has launched Sitecore Stream, an AI-powered brand assistant designed to let marketers interact with brand assets conversationally and visually. The goal: real-time creative collaboration with a system that understands a brand's tone and content.
Building the tool required evaluating available AI frameworks. Sitecore's search narrowed the field to options like LangChain, Semantic Kernel, and Prompt Flow, but found those platforms weighted toward backend processing.
“While these focused heavily on backend AI capabilities, they didn't emphasize the frontend experience—I like to use the term 'UI for AI.' In contrast, the AI SDK by Vercel offered a seamless integration of AI with front-end user experience.”
— Mo Cherif, Senior Director Generative AI, Sitecore
The Technical Stack Behind the Assistant
Sitecore's assistant combines a Python backend with a Next.js frontend, connected by the AI SDK. Streaming is central to the experience, delivering responses in real time rather than forcing users to wait for full payloads.
Pre-built SDK features shaped the development approach. The useChat hook simplified embedding chat AI into the design layout, managing input state, messages, loading indicators, and errors. Object generation capabilities let the team render structured output from the AI provider. These utilities let Sitecore's team sidestep low-level AI plumbing and focus on brand-specific interactions.
The integration also managed the flow of data between Azure-hosted backends and Next.js frontends. Server-side rendering and streaming helped the platform handle high data volumes while keeping responses fast enough for a conversational interface.
A Creative Partner, Not Just a Chatbot
Marketers using the assistant can ask for visual asset suggestions or content aligned with the brand's voice. The system replies with text, images, and content previews, creating what Sitecore describes as a virtual creative partner experience. This relies on agent-based models that carry a customer's brand knowledge and tone, presenting AI functions in a UI that requires no technical background.
Scaling to a Modular AI Future
Sitecore sees Stream as the foundation for more modular offerings. Features like brainstorming and asset discovery demand distinct workflows and user experiences. With the AI SDK's ability to render interactive widgets on the server and push them to the client, Sitecore plans to tailor these experiences per use case.
The choice positions the company to unfold new brand interaction features without rearchitecting the core integration.
