Ordering groceries via AI assistant: The 'headless' era is near

The era of interacting with AI assistants is rapidly evolving, moving beyond basic commands to sophisticated task execution. Imagine asking an AI assistant to find a recipe, then having it automatically search for ingredients online, add them to a shopping cart on a delivery service website, and await your final approval for purchase – all without human intervention. This is the reality demonstrated by OpenAI's "ChatGPT Agent," launched for paid subscribers in the US last July.

This advancement heralds the arrival of the "headless" era, where the user interface (UI) becomes increasingly detached from the core service operations. In this model, AI assistants can directly access and manipulate data across various applications and websites, performing tasks like shopping, booking, and managing schedules. This diminishes the need for users to manually navigate individual apps or websites.

“

Why should users have to log into Salesforce?

Global IT giants are already integrating this headless approach into their strategies. Salesforce, for instance, recently unveiled new services incorporating AI assistants into Slack, allowing users to access customer information and contract details directly within the messaging platform. Salesforce co-founder Parker Harris questioned the necessity of users logging into Salesforce directly, suggesting AI assistants will streamline access through familiar interfaces like Slack, with Salesforce's backend services acting as the "engine."

Technological enablers like Anthropic's "Model Context Protocol" (MCP), released in November 2024, are crucial to this transition. MCP standardizes how AI models access external data and services, enabling AI assistants to execute functions across diverse platforms within a single conversational flow. However, the widespread adoption of headless services faces resistance. Companies worry about the erosion of their existing revenue models, particularly those reliant on direct website traffic and advertising. If AI assistants handle all searches and transactions, users may visit individual service sites less frequently, impacting platform businesses.

“

As AI agents become more widespread, we anticipate that the existing app-centric service structure will evolve into a headless form where functions and interfaces are separated, and the role of the front-end will be reduced.