Discovery
A single finding about how something really behaves.
AI Mode Tags: FollowUp, Generate, Layout, Entity List and @prompt_leaks.ignore
Explore the core AI mode tags used
Your 'Gmail' can write code.
Gemini in Gmail drafted Python code to fine-tune a DeBERTa model for sentiment classification, complete with data loading, training loops, and evaluation.
Google, Anthropic and OpenAI crawlers fetched dejan.ai's robots.txt 1,150 times and its llms.txt 4 times in 30 days
In a 30-day server-log census of dejan.ai's three machine files, the big-three AI crawlers fetched robots.txt 1,150 times combined, llms.txt 4 times, and the OKF content bundle 23 times.
Finding Bard Inside Google's Gemma 4
A mechanistic interpretability study shows how ablating specific MLP neurons in Gemma 4 causes the model to revert to its older, dormant Bard identity.
Gemini picks whichever web page it reads first 92% of the time. Only pages rewritten to its measured preferences ever beat that.
Shown two web pages, Gemini picks the first one 92% of the time. Pages rewritten to 31 measured text preferences win 15% from the second slot. Ordinary pages: 0%.
The Paradox of Average and AI Visibility
A study by DEJAN AI examines how AI rankers prioritize prototypical content over brand authority, suggesting a dual strategy for SEO and human conversion.
Cascading Hallucination
A novel failure mode in multi-agent AI systems: a capable model trusts its sub-agents to deliver grounded facts, but they hallucinate, and the orchestrator launders the fiction into a confident answer without ever questioning it.
Google Uses Chrome to Supply Context to Gemini Chat
An analysis of Chrome's Page Content Agent, which builds a structured tree of page elements to provide Gemini with grounded context for reading and interaction.
How Search Grounding Biased an LLM Against YouTube
An analysis of how Claude's webinar platform recommendations were influenced by affiliate-driven content, and a correction regarding YouTube's live features.
Google’s (still) doesn’t see your live page.
Australian AI SEO agency specialising in brand visibility optimisation for global brands and e-commerce websites using machine learning and data science.
Search Grounding is Transient
Google’s AI search and Gemini use a single-turn transient architecture that purges raw web snippets from working memory immediately after a response is sent.
Implicit Queries in AI Search
An analysis of Google patent US11769017B1, detailing a system that uses context and implied input engines to proactively generate and push AI summaries.
Google’s AI Uses Schema?
An investigation into whether Google uses structured data to ground Gemini in AI search, exploring the relationship between LD+JSON and RAG grounding sources.
Google AI Search Update: Completely New Grounding Format
An observation of a new, custom grounding context format for Gemini that deviates from the traditional index-based model used in previous prompt types.
AI Mode, Content & Search Index
Tests suggest Google’s AI Mode uses a proprietary content store rather than retrieving live web content from the search index during the query fan out process.
How user prompts shape your content visibility in AI search.
An analysis of how AI search rankers use semantic alignment to surface different content zones within a single article based on query specificity and intent.
Gemini 3 hallucinates fan-out queries
An analysis of Gemini 3 API responses reveals the model fabricating search queries to justify its answers, demonstrating persistent hallucination behaviors.
AI Overviews = Dialogflow Agent?
An analysis of AI Overview leaks suggesting that Google's implementation may be based on the Dialogflow agentic framework, specifically regarding intent priority.
There’s a small army of on-device models coming to Chrome
Technical interpretations and parameter breakdowns for various AI models, including Gemini, Gemma, ULM, and StableLM, covering architecture and scale.
AI Mode Site Search
Explore Vertex AI website search features, including Enterprise edition tools like extractive answers, image search, and advanced LLM capabilities for summaries.
AI Mode & Page Indexing
Tests indicate Google's AI Mode uses a proprietary content store rather than the live web, as it fails to fetch indexed pages that are otherwise ranking.
AI Mode is Not Live Web
An experiment testing Google's AI Mode suggests it may rely on Google's existing index or cached web data rather than performing live HTTP requests for all URLs.
How AI Mode Selects Snippets
An analysis of how Google selects content for AI Mode snippets, identifying patterns in value propositions, HTML structure, and semantic selection criteria.
Chrome’s New Embedding Model: Smaller, Faster, Same Quality
Chrome's latest update features a new text embedding model that is 57% smaller than its predecessor, using int8 quantization to maintain search quality.