Context-Driven Interface Generation
- 2 minsA quiet change that AI agents are bringing is how we think about user interfaces. The traditiona old concept of web mashups might make a comeback with AI Agents. When agents interact with APIs dynamically (with no a priori knowledge of the subset of APIs they will use), something more fluid emerges: interfaces that generate themselves based on what the agent discovers and decides to do!
These context-driven dynamic mashups are such that the experience isn’t predetermined by frontend code but composed in real-time as agents interact with services.
At Ericsson Research, together with Vonage, we’ve been exploring how Model Context Protocol (MCP) enables agents to access both Communications Platform as a Service (CPaaS) APIs and Network APIs. MCP is an open standard that lets AI systems discover and use external tools without manual integration.
The technical implementation is straightforward: wrap APIs as MCP tools with defined schemas. For example, Vonage’s SMS API becomes a send_sms tool. The agent sees it’s available, understands its purpose from the schema, and can invoke it when needed. No hardcoded integration, no custom prompts—just dynamic capability discovery. Same thing goes for device_location, contact_information and others.
Demo: Agent dynamically composing Vonage APIs and network capabilities
But agents don’t just call APIs. They also compose the user interface, when a user asked about device location, the agent:
- Called the Device Location API to get coordinates
- Recognized coordinates alone weren’t useful
- Automatically crafted the HTML/CSS/JS bits to visualize the location
- Added the embedded map as the response in the chat
No developer wrote “if location query, then show map”. The agent made that compositional decision based on available tools and context. The UX was generated from the interface descriptions and the user intents.
In a future where you have agentic browsers like Atlas we will see more interfaces that assemble themselves based on runtime context, available services, and agent reasoning rather than static frontend templates and side-chatbots.