The website of the future may assemble itself for every visitor

2024-12-27 · Source: Latent.Space - Www.latent.space · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering, E-commerce & Digital Commerce · Depth: Intermediate, short

Summary

Adobe Principal Scientist Carlos Sanchez demonstrated the concept of an "agentic site" at the AI Engineer World's Fair, proposing a web experience that dynamically assembles personalized pages for each visitor. This "audience of one" approach interprets user intent from browsing behavior and search queries, then uses a Large Language Model (LLM) to retrieve and compose relevant content from a company's existing corpus in real time. Sanchez emphasized that this technology is feasible now, estimating an inference cost of "one to two cents per page" and a target generation time of "one to two seconds." While not yet broadly deployed, Adobe is exploring use cases like e-commerce and acknowledges the challenge of identifying optimal applications. The future web will likely need to cater to both human users and autonomous AI agents, requiring sites to support various interaction patterns and levels of delegation, potentially through structured tools like WebMCP.

Key takeaway

For AI Product Managers evaluating future web strategies, Adobe's "agentic site" concept suggests moving beyond static personalization to real-time, intent-driven page assembly. Your teams should experiment with dynamic content systems that can serve an "audience of one" and prepare for interactions with both human users and autonomous AI agents. Prioritize model efficiency to meet latency targets and explore structured content approaches for agent-to-agent commerce.

Key insights

Agentic sites can dynamically assemble personalized web experiences in real time for an "audience of one" using existing content.

Principles

• Websites can be dynamic, intent-driven systems.
• Personalization extends to real-time page composition.
• Sites must support human and AI agent interactions.

Method

Interpret visitor intent from signals, categorize it, then use an LLM to retrieve and compose a personalized page from existing content, targeting 1-2 second generation time.

In practice

• Prioritize model speed for site generation.
• Target e-commerce for agentic site pilots.
• Implement WebMCP for agent-accessible tools.

Topics

• Agentic Sites
• Website Personalization
• Large Language Models
• AI Agents
• Web Content Management
• E-commerce AI

Best for: Product Manager, CTO, VP of Engineering/Data, AI Engineer, AI Product Manager, Director of AI/ML

Related on AIssential

• Microsoft unveils tools to build infrastructure for agentic web — The agentic web, driven by AI agents, is rapidly expanding and requires new commerce infrastructure.
• Selling AI to Platform Owners By Leaning into Website Personalization and Generative UI — AI-driven personalization and generative UI can significantly boost platform engagement and conversions.
• Winning in the era of autonomous AI shoppers — Autonomous AI agents are rapidly reshaping retail, demanding immediate foundational shifts from retailers to remain competitive.
• This Website is not for Humans — The internet is evolving to an "agent-first" design, simplifying agent-driven content publishing.
• I Audited the Top 100 SEO Agencies in America for AI Agent Readiness. 83% Had Never Even Started. — Websites must adapt their technical structure, not just content, for autonomous AI agent navigation and citation.
• The Virtual Roundtable: Multi-Agent Personas Simulating the Dynamics of Human Brainstorming — Multi-agent AI personas can simulate human brainstorming, deepening idea refinement through structured discussion but not expanding initial diversity.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Latent.Space - Www.latent.space.