Notes from the picks-and-shovels side of AI search.

Google AI Mode and ChatGPT read different indexes (Google vs Bing) and cite almost nothing in common: 6.2% source overlap, 14 vs 3.7 sources per answer. The data, and how to get cited in each.

Google AI Mode and Gemini run on the same model but are different surfaces. What separates them, how each cites sources, their reach (AI Mode 1B vs Gemini app 950M), and which to optimize for.

AI shopping agents pick products by parsing structured data, not brand or design. How they evaluate options, the five signals they rank on, and the checklist to make your catalog agent-ready.

Agentic commerce is AI agents researching and buying for you. The three-layer stack, the ACP and UCP protocols, the payment rails from Visa and Mastercard, and what brands must fix to get picked.

ChatGPT draws on four sources: training data frozen at a cutoff, the live web via Bing, licensed publisher content, and your own context. What each is, and which one you can actually influence.

Copilot runs on the Bing index, so getting cited by Copilot is Bing SEO. ChatGPT pulls from the open web. The reach data (20M seats, 90% of the Fortune 500), the citation concentration, and how to get named in each.

Gemini takes 52% of its citations from brand-owned sites; ChatGPT takes 49% from third-party directories. Reddit is 0.1% of Gemini citations. The reach data, the citation split, and the two playbooks.

ChatGPT shopping recommends products by relevance, not ads. How it picks products, where its data comes from (83% mirrors Google Shopping organic), and the five steps to get your products recommended.

Perplexity does not rank pages, it cites sources (about 8.2 per answer). How Perplexity picks sources, the five ways to get cited, the domains and formats it favors, and how to allow PerplexityBot.

GPTBot is OpenAI's training crawler. Blocking it opts you out of AI training, not ChatGPT search, which runs through OAI-SearchBot. The four OpenAI bots, the robots.txt config most brands want, and how to verify GPTBot by IP.

llms.txt is a Markdown file that tells AI models which pages matter. What it is, how to create one, how it differs from robots.txt, and the honest verdict on whether it works in 2026.

Perplexity cites brands over 20x more than ChatGPT (13.05% vs 0.59%), but ChatGPT reaches far more people and only 11% of their sources overlap. The data, and how to win each.

Otterly, Peec, Profound, Semrush, and the API option compared by coverage, alerts, and price. The best AI monitoring tools and rank trackers for brands, agencies, and developers.

Google rank tracking is solved; AI search tracking is not. What to measure, how to track across ChatGPT, Perplexity, Gemini, and AI Overviews, and the tools and APIs that scale it.

Your competitor might be ChatGPT's default pick in your category. How to track rival mentions and citations across every AI engine, benchmark share of voice, and pick the right tool.

AI Overviews change daily and sit above your links. Compare AI Overview tracking tools, what to look for, dashboards vs APIs, and how to monitor whether Google cites your brand.

Ranking #1 is no longer the finish line. AI search optimization is how you get cited inside AI answers across every engine: what it is, how engines pick sources, the metrics, and the tools.

LLM SEO tools track whether ChatGPT, Gemini, Claude, and Perplexity mention and cite your brand. The three types, what to look for, dashboards vs API, and how to start.

LLM SEO (LLMO) is optimizing content so ChatGPT, Gemini, and Claude understand, trust, and cite it. What it is, how it differs from SEO, the metrics, and how to start.

The repeatable six-step GEO loop: baseline your citations, fix AI crawlability, structure content for retrieval, add citable data, keep it fresh, and measure share of voice.

AEO tools do one of two jobs: track whether AI answers cite you, or optimize content to become the answer. The best trackers, optimizers, and the API option.

A reference of ChatGPT knowledge cutoff dates by model, what a cutoff is, how browsing extends it, and why it decides whether ChatGPT knows your brand exists.

ChatGPT searches the web before it answers, but cites only 15% of the pages it reads. How retrieval and citation work, what gets cited, and how to show up.

No, and the reason matters for your brand. Why ChatGPT answers vary by user, settings, and time, even at temperature 0, and why one screenshot tells you nothing.

Answer engine optimization vs generative engine optimization: how they differ, where they overlap, whether they are the same thing, and which to prioritize.

SEO got you found; AEO gets you quoted. What answer engine optimization is, how it works, AEO vs SEO, AEO vs GEO, and how to start getting cited in AI answers.

The GEO market split into tools, agencies, and APIs. A comparison of the best GEO platforms, agencies, and the API option, by what they do and who they fit.

Search used to mean a spot in the ten blue links. Now it means being one of the few sources an AI cites. What GEO is, how it works, GEO vs SEO, and how to start.

See how ChatGPT, Perplexity, and Gemini describe your brand. A free, step-by-step AI visibility audit you can run in 20 minutes, plus the best free checker tools.

You can rank #1 on Google and still be invisible in ChatGPT. What AI visibility means, how it differs from SEO, and how to measure it across every engine.

Profound, Peec, Otterly, Ahrefs, Semrush, and the API option compared by coverage, price, and who they fit. A dozen tools, $300M+ in funding, prices from $29 to $499/mo.

The AI answer box now sits above the blue links on a huge share of searches. What Google AI Overviews are, how they work, what they do to traffic, and why being cited matters.

Ranking #1 no longer wins the answer. The 2026 playbook for getting cited in Google AI Overviews: self-contained passages, schema, entity authority, and how to track your citations.

ChatGPT, Perplexity, Google AI Mode, Gemini, Copilot, and Grok ranked by accuracy, citations, and use case. Only 11% of cited sources overlap between engines, so the right pick depends on the job.

The complete build of a Generative Engine Optimization tool. Four atoms, six surfaces, daily cadence, real cost numbers. One weekend of work, ~$120 a month to run.

Track which URLs AI engines cite as answers. Citation extraction, gap analysis, and an outreach feedback loop. Around 150 lines and ~$50 a month.

Build the single visibility score every CMO asks for. One number, plotted over time, sourced from real AI answers across six surfaces. ~80 lines, $30 a month.

Track brand position inside ChatGPT, Claude, Gemini, and Perplexity answers the same way you track Google rankings. ~100 lines, $15 a month.

Compare how ChatGPT, Claude, Gemini, and Perplexity treat your brand differently. Per-model gap reports drive specific optimizations the rollup score hides.

Build an AEO and GEO automation agent on Hermes in one evening. Five steps, real config, $15 to $30 per month to run.

Six concrete automations you can run on Hermes today: visibility checks, citation gap analysis, share-of-voice tracking, AI Overview alarms, prompt discovery, and content evals.

Wire MentionsAPI into Hermes Agent and watch every ChatGPT, Claude, Gemini, and Perplexity answer about your brand. Setup in 10 minutes, from 5 credits per check.

A 1,000-query teardown of how badly the LLM APIs you are optimizing for diverge from what real users actually see in the UI. With numbers, sources, and the picks-and-shovels framing.