AutoGEO: An Automatic Optimization Framework for Generative Search Engines

The CMU team proposed the AutoGEO (Generative Engine Optimization) framework, which uses machine learning to automatically learn the preferences of generative engines, enables intelligent rewriting of web content, enhances visibility and traffic acquisition capabilities in AI searches, and helps content creators address the new SEO challenges in the era of generative search.

Traditional SEO focuses on keyword matching, backlinks, and page structure. However, the rise of generative AI search tools like ChatGPT and Perplexity has changed how users access information—they directly generate comprehensive answers instead of returning link lists. This poses a new challenge for content creators: how to make their content "seen" and cited by AI?

AutoGEO is an innovative framework proposed by the CMU team. Unlike traditional SEO that relies on manual experience, it uses machine learning to automatically learn the preference patterns of generative engines. Its core insight is: when building answers, generative engines select citation sources based on dimensions such as content structure, information density, and credibility. AutoGEO reverse-engineers these preferences through experimental design and feedback learning, converting them into executable optimization strategies.

AutoGEO consists of three key components: 1. Preference Learning Module: Interacts with generative engines to collect citation data of content variants and builds a preference probability model; 2. Content Rewriting Engine: Performs semantic reconstruction based on the preference model (adjusting argument order, enhancing data support, etc.); 3. Effect Verification Loop: Re-submits optimized content for evaluation and iteratively adapts to changes in engine algorithms.

Application scenarios include corporate official websites/blogs, news media, e-commerce product pages, academic institutions, etc., helping to increase content exposure, citation rate, or recommendation level in AI searches. For creators, it represents a paradigm shift from "writing for search engines" to "writing for generative AI", requiring a rethinking of how to become a cited authoritative source.

AutoGEO marks SEO entering a new stage, where optimization for AI search will become a standard in digital marketing. The open-source release based on ICLR 2026 research results provides a foundation for the industry. In the future, we can expect directions such as vertical domain-specific models, multi-language preference adaptation, and deep integration with content management systems.

Generative search engines are reshaping the information distribution landscape, and AutoGEO provides a scientific and systematic response method. For creators and organizations that want to maintain content competitiveness in the AI era, understanding and applying such tools is a key capability, and data-driven optimization can effectively improve content performance in generative engines.