Perplexity GEO: Photon Replaces Its Third-Party Retrieval Engine
Perplexity says its in-house Photon engine cut retrieval and ranking p99 latency from about 800 ms to 65 ms, and shipped a Fast Search tier at $1.00 per 1,000 requests.
From the author
Recent Articles From Julian Mercer
Search Console Now Reports Multimodal Search: What Image-Based Queries Reveal
Google added a multimodal search type to Search Console on September 24, 2026. Here is what it covers, what it does not, and a 30-minute audit to run on your own data.
Laya for Internal Linking, and Where It Breaks
Internal linking is thousands of yes/no decisions, which is exactly what a decision model is for. It is also where Laya's wide-option weakness bites hardest. Here is the workflow, the limit, and the cascade that works around it.
Laya as a Local Reranker for Internal Search and RAG
Similarity search finds passages that look related. It does not find passages that actually answer the question. Here is how to add a local Laya reranking stage, how to measure whether it helped, and where it fits in SEO work.
Running Laya Locally: Hardware, Latency, and What Free Actually Costs
Laya has no per-call cost, but self-hosting is not free. Here is the real cost model: latency by hardware tier, throughput planning, a break-even calculation, and the point where a hosted API is still the better choice.
We Rebuilt a 566-Page Internal Link Map with Jev. Only 287 Links Survived Review.
A Jev internal-link run proposed 679 links across 566 pages in under six seconds for 27 cents. An editorial pass kept 287 of them. Here is the full attrition, the rubric rewrite that improved recall from 45% to 65%, and how to run the same benchmark on your own site.
How to Set Up a Grok Bot SEO Agent (Step-by-Step)
A complete walkthrough for building one Grok Bot that audits your site, reads Search Console, and reports back, without giving it permission to change anything until you have verified it.
Which Data Sources Should You Care About For AI Agents?
A tiered map of the data sources AI answers actually pull from, plus the API or MCP path that lets Codex, Claude Code, Hermes Agent, OpenClaw, Pi Agent, Grok Bot, and Meta Muse read each one directly.
How One Team Built a Jev Internal-Link Analyzer for 78 Yen
A real build teardown: how a Jev-powered internal-link analyzer crawls a site, narrows candidates with TF-IDF, extracts the exact sentence to link from, and proposes anchor text that already exists on the page — for about the price of a coffee.
How to Build a Jev Internal-Linking and Cannibalization Workflow
A runnable pipeline that turns a crawl export into a reviewed internal-link and merge worklist. Two-stage candidate selection, exact Jev question schemas, confidence routing, and the checks that keep you from shipping bad links at scale.
How to Check Domain Availability and Registration Price With Codex (Full SKILL.md Included)
A Codex skill that splits one domain question into three: free registry state from RDAP, free Cloudflare prices from a public socket, and the token-gated check.
How to Set Up Cloudflare's AI Crawler Policy After the September 15 Default Change
Cloudflare split AI bot traffic into Search, Agent, and Training on September 15, 2026, and ad-supported sites now get Training and Agent blocked by default. Here's how to check where you landed and configure each category on purpose.
How to Tell Which AI Agents Are Crawling Your Site (and Which to Block)
A repeatable way to identify the AI agents hitting your site, separate verified search crawlers from unverified agent traffic, and set access rules you can defend.
Google Search Console MCP: What Four SEO MCP Servers Actually Expose
We connected four SEO MCP servers and asked each one for its tool list. They returned 42, 21, 4, and 1. That spread is the whole decision: a wrapper, a gateway, or a single gimmick.
How to Use Copilot in Bing Webmaster Tools: A Practical 2026 Walkthrough
Copilot in Bing Webmaster Tools answers questions about your site data in plain language. Here is a step-by-step 2026 workflow for asking the right questions, verifying every answer, and turning it into a weekly review habit.
Google Said It Can Lose Faith in Your Whole Site: What Mueller Actually Told One Publisher
Google's John Mueller told a publisher on 7 September 2026 that its systems may have lost faith in his site based on old programmatic pages. The full thread is more specific, and more useful, than the quote going around.
Google AI Mode Tests Paginated Follow-Up Responses With Skip Button
Google is testing paginated follow-up responses in AI Mode with a skip button, changing how conversational AI search presents multi-turn interactions.
Google's John Mueller Warns Programmatic SEO Can Cause Sites to Lose Search Trust
Google's John Mueller warned that programmatic SEO strategies can cause Google's systems to lose faith in an entire site, with recovery timelines comparable to spam penalties.
Apple Clarifies Applebot-Extended Rules Do Not Affect Search Rankings
Apple updated its Applebot documentation to explicitly state that Applebot-Extended directives do not affect search rankings, clarifying the separation between AI training opt-outs and search visibility ahead of iOS 27.
ChatGPT Runs Its Own Search Index Family Called Labrador, Research Reveals
OpenAI's ChatGPT operates a multi-layered retrieval system internally called Labrador — a family of at least 12 specialized indexes covering general web, PDF, YouTube, news, arXiv, Wikipedia, local businesses, finance, legal, medical, shopping, and images.
Google Ads Tests Serving Exact and Phrase Match Campaigns in AI Mode
Google confirms it is testing serving regular Search campaigns with exact and phrase match keywords in AI Mode, expanding ad eligibility beyond AI Max and Performance Max.
How to Disavow Toxic Backlinks in 2026: Google's Disavow Tool, Step by Step
Google's John Mueller said in September 2026 that the disavow tool does have effects. This step-by-step guide covers when disavowing is worth it, how to build a valid disavow file, and how long to wait before you judge the result.
Google Ads Adds Pre-Fill for Asset Groups Using Asset Studio Generated Assets
Google Ads updated its Performance Max documentation to describe a workflow that lets advertisers pre-fill asset groups with AI-generated assets created in Asset Studio.
