CapstonAI is an AI visibility platform that helps brands track, audit, and optimize their presence in AI-generated answers. It monitors how your brand is mentioned across ChatGPT, Claude, Gemini, Perplexity, and Google AI Overviews, then provides actionable fixes to improve your share of voice in AI search results.

How does CapstonAI track AI visibility?

CapstonAI sends structured brand queries to multiple large language models (LLMs) on a recurring basis and records whether your brand is mentioned, cited, or recommended. It aggregates these results into a GEO score, a share-of-voice metric, and a citation report.

Which AI models does CapstonAI monitor?

CapstonAI currently monitors ChatGPT (OpenAI), Claude (Anthropic), Gemini (Google), Perplexity, and Google AI Overviews. Coverage is continuously expanded as new AI search surfaces emerge.

CapstonAI offers a free WordPress plugin that provides basic AI visibility scanning for your site. For full AI visibility tracking, audit reports, automated content fixes, and multi-brand monitoring, the Core Plan is available at $99/month.

How is CapstonAI different from Semrush or Ahrefs?

Semrush and Ahrefs are built for traditional search engine optimization. CapstonAI is purpose-built for the AI search era: it tracks your brand presence inside AI-generated answers from ChatGPT, Claude, Gemini, and Perplexity.

The 5 Pillars Of Generative Engine Optimization: Strategic Agenda From Toronto Empirical Study

The 5 Pillars of Generative Engine Optimization: Strategic Agenda from Toronto Empirical Study

The University of Toronto study by Chen et al. (arXiv:2509.08919, 2025) closes with a strategic GEO agenda derived from their empirical findings across multiple verticals, languages, and engines. The agenda is built on 5 strategic pillars: (1) engineer for agency and scannability, (2) dominate earned media across all engines, (3) adopt engine-specific and language-aware strategies, (4) overcome the big brand bias for niche players, (5) optimize content strategy for justification and comparison (winning the shortlist). Below: each pillar, the empirical evidence behind it, and the operational playbook to implement it.

TL;DR: The 5 GEO pillars from Chen et al. are: technical fundamentals + schema (treat your site as an API for AI), earned-media dominance (the #1 driver of AI citations), engine + language-specific tactics, niche-brand differentiation strategies, and justification-rich comparison content. The first 3 pillars are foundational; the last 2 separate winners from losers within a category.

Free CapstonAI scan → GEO Research hub

Pillar 1: Engineer for agency and scannability

The Toronto study emphasizes that AI engines function as agents that parse, interpret, and synthesize information. Websites cluttered with marketing fluff and unstructured content fail. The recommendation: treat your website as an API for AI. This requires rigorous technical SEO fundamentals plus detailed Schema.org markup for all entities — products, specifications, prices, reviews, warranty details, availability.

Operational playbook:

Deploy FAQPage, HowTo, Product, Organization, BreadcrumbList schema across all priority pages.
Maintain a clean robots.txt and llms.txt that allows GPTBot, PerplexityBot, ClaudeBot, GoogleBot.
Audit Core Web Vitals (LCP, INP, CLS) quarterly. Pages that fail technical performance get filtered out of candidate pools.
Use semantic HTML (proper h1/h2/h3 hierarchy, ul/ol lists, table elements for tabular data).
Embed entity disambiguation: sameAs links in Organization schema connecting to Wikipedia, Wikidata, Crunchbase, LinkedIn.

Pillar 2: Dominate earned media across all engines

The most consistent and significant finding in the Toronto study is the overwhelming bias of AI engines toward earned media. Claude and ChatGPT are extremely earned-heavy (above 80% in most verticals). Perplexity and Gemini incorporate more brand and social content but still prioritize earned sources. To win in AI search, shift focus from creating owned content to systematically earning third-party validation.

Operational playbook:

Build a tier-1 PR program targeting authoritative publications in your vertical (Reuters, Forbes, vertical-specific media).
Earn placements in independent review sites that AI engines cite repeatedly (RTINGS, CNET, Wirecutter, Consumer Reports).
Pursue Wikipedia notability where eligible — encyclopedia sources dominate absorption in the Zhang et al. (2026) framework.
Invest in podcast outreach with transcript SEO so transcripts are crawlable and citable.
Build profiles on aggregators relevant to your vertical (G2, Capterra for SaaS; Healthgrades for healthcare; OpenTable for restaurants).

Pillar 3: Adopt engine-specific and language-aware strategies

The Toronto data shows engine-specific patterns are sharp: Claude is English-stable across languages; GPT swaps entire site ecosystems per language; Perplexity uniquely cites YouTube; Gemini leans more brand-friendly than Claude or ChatGPT. Pairwise cross-engine Jaccard is 0.10-0.25. Treating “AI Search” as a monolith wastes budget.

Operational playbook:

Run engine-specific prompt panels: ChatGPT, Perplexity, Claude, Gemini each get their own 30-prompt panel.
For Perplexity: invest in YouTube content and Reddit community presence.
For Claude: maintain English Wikipedia presence; English authority coverage compounds across all language markets.
For GPT: build local-language earned coverage per target market (near-zero cross-language overlap).
For Gemini: optimize brand-owned content because Gemini is the most brand-leaning engine in the data.
Build engine-specific dashboards tracking citation share, source mix, and Jaccard overlap with top competitors.

Pillar 4: Overcome the big brand bias for niche players

The Toronto cola experiment showed that unbranded prompts default to market leaders: 62.2% major brands, 9.0% niche brands, 28.8% other (combined across ChatGPT and Perplexity). Niche brands must over-invest in building tangible, verifiable authority in narrow niches rather than competing on broad category-level prompts.

Operational playbook:

Define 5-10 narrow, dominable niches where your brand has credible top-3 positioning (not broad categories where market leaders win).
Build deep expert content in your niche: comparison guides, definition pages, procedural how-to content.
Invest in YouTube reviews and Reddit community presence for Perplexity visibility.
Earn placements in specialty vertical press, not generic publications.
Track niche-brand citation share within your defined narrow category as the realistic KPI (not total category share).

Pillar 5: Content strategy — justify and compare for the shortlist

The low domain overlap between engines indicates each AI synthesizes from different sources, but all seek clear unambiguous justification. AI search is not about generating ten blue links but about justifying a placement on a synthesized shortlist. Brands must engineer content explicitly to answer comparison questions and provide extractable reasons for superiority.

Operational playbook:

Build comparison tables: “Brand X vs Brand Y vs Brand Z” with criteria-by-criteria rows (price, features, support, integrations).
Use bulleted pros-and-cons lists for every product page.
Bold value-proposition statements (“longest battery life”, “best for small families”, “lowest total cost of ownership”).
Include definition markers (“FAQPage” schema, glossary entries, “what is X” sections) per the Zhang et al. (2026) +57% influence finding.
Include quantitative evidence: numerical specs, cohort data, benchmarks. The Zhang et al. evidence-genre data shows numbers/statistics drive +62% influence uplift.

The integrated 8-week GEO launch checklist

Week	Pillar focus	Deliverable
1	Pillar 1 + audit	Schema deployment audit, llms.txt/robots.txt review, baseline prompt-panel run
2	Pillar 3	Engine-specific dashboards built (4 engines, weekly tracking)
3-4	Pillar 2	PR program launch + first 3 placements pitched; Wikipedia eligibility analysis
5	Pillar 5	Top 10 comparison pages published with full schema, tables, definitions
6	Pillar 4 (if niche)	Narrow niche positioning + YouTube + Reddit strategy live
7-8	All pillars	Quarterly review cadence established; CFO/CMO reporting template deployed

Common errors when applying the 5 pillars

Doing all 5 pillars at 30% effort. Concentrated execution on 2-3 pillars beats distributed effort across 5. Pick the 2 pillars with the highest leverage for your stage.
Treating Pillar 1 as optional. Schema, technical SEO, llms.txt are the gate. Without them, the other 4 pillars cannot perform.
Skipping engine-specific dashboards. Without measurement per engine, you cannot tell whether your Pillar 2 PR investment is hitting Claude, ChatGPT, or both.
Spending Pillar 5 effort on brand-owned promotional pages. Comparison content must be honest and bidirectional. Pages that only argue your brand wins lose trust signals.
Ignoring Pillar 4 for major brands. Major brands also benefit from niche positioning — just at a different scale. Define narrow niches even if you dominate the broad category.

FAQ — The 5 pillars of GEO

Which pillar has the highest ROI?

Pillar 2 (earned media) in the CapstonAI partner cohort Q1 2026: median +28% citation share within 90 days for accounts that launched a tier-1 PR program. Pillar 1 (technical/schema) is foundational but doesn’t move the needle alone. Pillars 3, 4, 5 are differentiators within a category.

How long until the 5 pillars compound?

Pillar 1 effects visible within 30 days (schema parsing, technical signal). Pillar 5 (comparison content) within 60-90 days. Pillar 2 (earned media) 90-120 days. Pillar 3 (engine-specific tactics) ongoing optimization. Pillar 4 (niche differentiation) typically 6-9 months for full compounding.

Can a small team execute all 5 pillars?

Realistically, a 1-2 person team can run Pillars 1, 3, 5 in-house and outsource Pillar 2 (PR agency). Pillar 4 requires founder-led strategic positioning, not headcount. Larger teams (5+) can run all 5 pillars in-house.

Tools and related reading

Ready to implement the 5 GEO pillars?

Free CapstonAI scan →

Last updated: May 2026. Primary source: Chen, M., Wang, X., Chen, K., & Koudas, N. (2025). Generative Engine Optimization: How to Dominate AI Search. University of Toronto. arXiv:2509.08919. https://arxiv.org/abs/2509.08919