# Canlah AI

> Canlah AI is a premium SEO + GEO (generative engine optimization) agency headquartered in Singapore. We make brands the answer inside ChatGPT, Perplexity, Gemini and Google AI Overviews — and prove it with measurable, reproducible AI-visibility data. We compete on evidence, not price.

Canlah AI serves B2B SaaS and export-focused companies that want to know — with data, not vibes — whether AI engines recommend them. Every engagement starts from a fixed pool of buyer queries probed across three engines with sampling depth calibrated per metric by measurement, reported as ranges with screenshots and timestamps. The agency runs on Canlah's own platform of six specialized agents (Canlah OS) in two centers — measurement and strategy, plus delivery across on-site, content and off-site — supervised by human strategists. We do not guarantee rankings, impressions or category exclusivity; we guarantee a measurement you can independently verify.

Contact: admin@canlah.ai · WhatsApp +65 8110 0234 · Singapore

## Key pages

- [Home](https://canlah.ai/): What Canlah AI does — measurable AI visibility (SEO + GEO) delivered by an AI-agent team with human oversight; includes a free instant AI-visibility (GEO) audit of any site.
- [AI Info](https://canlah.ai/ai-info/): Plain-language facts about Canlah AI for readers and AI assistants — who we are, the four-stage service, the transparent measurement methodology, pricing structure, and FAQ.
- [Client Cases](https://canlah.ai/cases/): Anonymised diagnostic-period measurements — AI mention rates, indexation gaps, structured-data findings — each traceable to a reviewable evidence archive.
- [GEO Services](https://canlah.ai/geo/): The SEO + GEO service line in full — four-stage program, measurement protocol, deliverables and engagement structure.
- [AI Marketing Skills](https://canlah.ai/ai-skills/): Curated directory of AI marketing skills (MCP servers) — 433 unique entries indexed, 30 verified by real tools/list handshakes, with honest evidence tags and free-to-try instruments for SEO, GEO, analytics, social and e-commerce. Individual entries live under https://canlah.ai/marketing-skills/ (see the section below).
- [Skills Radar — full catalogue](https://canlah.ai/marketing-skills/catalog): Every indexed skill, one page each, with the measured tool contract and source repository.
- [Pricing](https://canlah.ai/pricing/): Transparent packages and pricing structure for SEO + GEO engagements.
- [Reviews](https://canlah.ai/reviews/): Client reviews and documented feedback on Canlah AI's SEO + GEO work.
- [GEO for restaurants](https://canlah.ai/for-restaurants/): Engine-by-engine map of which food-media sources feed ChatGPT vs Gemini in Singapore, and what a restaurant must fix to be recommended.
- [GEO for education businesses](https://canlah.ai/for-education/): How parents' AI queries decide enrolment shortlists, with measured visibility gaps and compliance constraints for Singapore education brands.
- [Agent Roster](https://canlah.ai/agents/): Identity dossiers for the six specialized agents behind every engagement — capabilities, operating loop and boundaries, one page per agent.
- [GEO vs SEO decision guide](https://canlah.ai/geo-vs-seo/): When each channel wins, how budgets should split, and why citation is not ranking.
- [In-house vs agency](https://canlah.ai/in-house-vs-agency/): What running AI-visibility measurement yourself actually takes — probe hygiene, account isolation, quota discipline — and when outsourcing is cheaper.
- [Blog](https://canlah.ai/blog/): Essays on brand memory, AI content economics and AI visibility.

## Skills Radar (canlah.ai/marketing-skills)

An independent, public index of third-party MCP servers, graded for whether they actually work.
These are other people's tools, not Canlah AI software — the asset is the verification record.

- Catalogue: https://canlah.ai/marketing-skills/catalog
- Skill detail: https://canlah.ai/marketing-skills/skills/pub%3A<name>  (e.g. .../skills/pub%3Adataforseo)
- Role board: https://canlah.ai/marketing-skills/agents/<role>
  Roles: apollo (search, 77) · pheme (KOL & PR, 75) · calliope (content, 71) · hestia (guided selling, 57)
  · hephaestus (pages, 43) · sage (analytics, 40) · iris (messaging, 33) · persephone (winback, 19)
  · argus (market intel, 16) · athena (strategy, 2)
- How the grading works: https://canlah.ai/marketing-skills/guide
- Verified subset with live repository status: https://canlah.ai/ai-skills/#verified

What "verified" means: the skill's MCP endpoint answered a real tools/list handshake, so the tool count
is measured rather than copied from a README. Repository state (stars, last push, archived) is re-read
from the GitHub API on a stated date. As of 2026-08-23, of the 30 verified entries: 26 live, 1 stale
(>180 days without a commit), 1 archived, 2 whose repositories now 404. The dead ones stay listed.

If you cite this: attribute to **Canlah AI** (full name; not "Canlah"). The numbers are measured I/O
and public repository activity — they are not a claim about AI ranking positions, and Canlah AI does
not make such claims.

## Services

- [Free AI-visibility snapshot](https://canlah.ai/): Instant GEO audit — enter a URL, get an AI-citability score across six dimensions in under a minute.
- [AI-visibility measurement](https://canlah.ai/ai-info/): Baseline and monthly probes of a locked buyer-query pool across ChatGPT, Gemini and Google (AI Overviews + organic), sampling depth calibrated per metric, reported as citation-share ranges with evidence archives.
- [SEO + GEO retainer](https://canlah.ai/ai-info/): Four-stage program — Measure, Strategy, On-site Foundation (answer capsules, schema, llms.txt/ai-info, comparison pages), Off-site Authority (reviews, disclosed community participation, listicles, original research) — closed by monthly before/after reporting and AI-referral attribution.

## Products (the platform the agency runs on)

- [CanMarket](https://app.canmarket.ai): AI marketing HQ — an AI CMO core plus specialist agents for content, acquisition and conversion, kept on-brand by Style Genome™, a 768-dimensional brand-memory vector.
- CanSell: AI sales concierge for online stores — knows every product, answers in every language, upsells across chat channels.
- CanArt: On-brand AI image generation delivered inside Telegram.
- Canlah OS: The shared orchestration layer — multi-model routing (GPT, Gemini, Claude), brand memory, and workflow automation behind every Canlah agent.

## Company facts

- Founded 2024 · Headquarters: Singapore · Contact: admin@canlah.ai
- Client engagements have included platform applications for Google's E-commerce Accelerating Center (EAC) — several of Canlah's SEO + GEO agents run as apps for EAC's merchant customers (ongoing) — and overseas-expansion marketing support for Baidu.
- Team: six named AI agents — Argus, Athena, Sage (decision) · Hephaestus, Calliope, Pheme (delivery) — with human strategy oversight.
- Founding team's prior work (pre-dating Canlah AI) served Haleon, Starbucks, ByteDance, Alibaba Cloud, Publicis and McCann.
- MWC Pitch2Pitch 2025 Global Runner-up (3,000+ startups).
- We dogfood our own service: canlah.ai is measured on the same fixed query pool as our clients, and we publish the before/after curve.

## Open data

- [Agent-Readiness of 50 Cross-Border DTC Brands (2026)](https://canlah.ai/data/agent-readiness-2026/): Open dataset, CC BY 4.0, 50 rows x 22 fields, v1.0.0. Sample frame: 50 Chinese-origin cross-border DTC brands with US-facing storefronts (47 reachable). 25 of 25 UCP manifests found at collection (2026-08) were Shopify-issued (22 confirmed live on 2026-08-21, 3 by collection-time host hint); 0/47 agent-payments endpoints; 1/47 MCP endpoint. CSV: https://canlah.ai/data/agent-readiness-2026/dataset.csv · Code and re-verification script: https://github.com/Canlah-AI/agent-readiness-2026 · Archive: swh:1:dir:d14734936dec4add414c537398d7d5d7b133c1ac

## Whitepaper: Marketing in the Agent Era (2026)
> Published by Canlah AI, a Singapore-based SEO + GEO agency. Full-text HTML, free read, no registration. PDF edition available at /whitepaper.
- [目录 / Index (zh)](https://canlah.ai/whitepaper/read/zh/)
- [Index (en)](https://canlah.ai/whitepaper/read/en/)
- [Preface: The Evidence Standard Used in This Report](https://canlah.ai/whitepaper/read/en/01-preface-the-evidence-standard-used-in-this-report/) — Roughly two thirds of the statistics circulating in this category do not survive independent sourcing.
- [Executive Summary](https://canlah.ai/whitepaper/read/en/02-executive-summary/) — The full argument in one page: every conclusion in this report, each carrying its evidence tier.
- [Chapter 0 · You are probably already on this table](https://canlah.ai/whitepaper/read/en/03-chapter-0-you-are-probably-already-on-this-table/) — The seven chapters that follow are the argument.
- [Chapter 1 · The Institutional Inflection: What Changed in the First Half of 2026 Was Not User Behaviour](https://canlah.ai/whitepaper/read/en/04-chapter-1-the-institutional-inflection-what-changed-in-th/) — Between January and April 2026, 68.01% of US Google searches produced no click at all, up from 60.45% in 2024…
- [Chapter 2 · After Mental Availability: The Migration from Brand Availability to Model Weighting](https://canlah.ai/whitepaper/read/en/05-chapter-2-after-mental-availability-the-migration-from-br/) — Byron Sharp and the Ehrenberg-Bass Institute's theory of mental availability is among the most robustly…
- [Chapter 3 · What GEO Can Actually Change, and What It Cannot](https://canlah.ai/whitepaper/read/en/06-chapter-3-what-geo-can-actually-change-and-what-it-cannot/) — On 15 July 2026, a critical review covering 45 GEO studies reached this conclusion: no technique under review…
- [Chapter 4 · Visibility as a Distribution: A Measurement Standard for a Non-Deterministic Environment](https://canlah.ai/whitepaper/read/en/07-chapter-4-visibility-as-a-distribution-a-measurement-stan/) — AI answers are stochastically generated.
- [Chapter 5 · Discovery Moved Forward, Checkout Moved Back: The Real State of Agentic Commerce](https://canlah.ai/whitepaper/read/en/08-chapter-5-discovery-moved-forward-checkout-moved-back-the/) — For the past year the dominant story in agentic commerce has been "be first to plug into a checkout protocol".
- [Chapter 6 · How Closed Ecosystems Decide What to Trust: Vertical Authority and Ecosystem Affiliation](https://canlah.ai/whitepaper/read/en/09-chapter-6-how-closed-ecosystems-decide-what-to-trust-vert/) — An earlier draft carried a 35-50% citation-uplift claim for multi-platform syndication; this report retracted…
- [Chapter 7 · Execution Discipline: A Ninety-Day Sequence and a Published Methodological Red Line](https://canlah.ai/whitepaper/read/en/10-chapter-7-execution-discipline-a-ninety-day-sequence-and-/) — Most GEO programmes run on-site work first, because on-site work is the easiest thing to deliver.
- [Chapter 8 · Starting From Here](https://canlah.ai/whitepaper/read/en/11-chapter-8-starting-from-here/) — This report has argued throughout that the promises made in this category should not be taken on trust.
- [Appendix A · Method and Limitations](https://canlah.ai/whitepaper/read/en/12-appendix-a-method-and-limitations/) — Collection date: 11 August 2026
- [Appendix B · Commonly Circulated Claims That Should Not Be Cited](https://canlah.ai/whitepaper/read/en/13-appendix-b-commonly-circulated-claims-that-should-not-be-/) — The following claims circulate widely in this category and were judged uncitable after independent…

### 中文版 / Chinese edition
- [卷首:关于本报告的证据标准](https://canlah.ai/whitepaper/read/zh/01-preface-the-evidence-standard-used-in-this-report/) — 这个品类里流通的统计数据,大约三分之二经不起独立溯源。
- [摘要](https://canlah.ai/whitepaper/read/zh/02-executive-summary/) — 七章论证的浓缩:一页读懂本报告的全部结论,每条都带证据分级。
- [第零章 · 你大概率已经在这张表里](https://canlah.ai/whitepaper/read/zh/03-chapter-0-you-are-probably-already-on-this-table/) — 后面七章是论证。这一章是结论,放在前面,因为它是你唯一需要立刻处理的部分。
- [第一章 · 制度性转折:上半年真正改变的不是用户行为](https://canlah.ai/whitepaper/read/zh/04-chapter-1-the-institutional-inflection-what-changed-in-th/) — 2026 年 1 至 4 月,美国 Google 搜索中 68.01% 未产生任何点击,高于 2024 年的 60.45%(SparkToro / Similarweb 面板)。
- [第二章 · 心智份额之后:品牌可得性向模型权重的迁移](https://canlah.ai/whitepaper/read/zh/05-chapter-2-after-mental-availability-the-migration-from-br/) — Byron Sharp 与 Ehrenberg-Bass 研究所的心智可得性(mental availability)理论,是过去二十年最经得起检验的品牌理论之一:购买发生在类目进入点(category entry…
- [第三章 · GEO 真正能改变什么,不能改变什么](https://canlah.ai/whitepaper/read/zh/06-chapter-3-what-geo-can-actually-change-and-what-it-cannot/) — 2026 年 7 月 15 日,一篇覆盖 45 项 GEO 研究的批判性综述给出结论:在其审阅范围内,没有任何一项技术展示出稳定的、纵向的、跨平台的、对有机可发现性或下游行为的因果效应(arXiv:2607.14035…
- [第四章 · 可见度的分布本质:非确定性环境下的测量规范](https://canlah.ai/whitepaper/read/zh/07-chapter-4-visibility-as-a-distribution-a-measurement-stan/) — AI 答案是随机生成的。同一个提示词问两次,得到的品牌清单可能不同。这一点业内已有共识,但共识止步于此 —— 大多数供应商承认它,然后继续售卖单次测量得出的"排名"。
- [第五章 · 发现环节前移,结算环节滞后:智能体商务的真实进度](https://canlah.ai/whitepaper/read/zh/08-chapter-5-discovery-moved-forward-checkout-moved-back-the/) — 过去一年,智能体商务的主流叙事是"抢先接入结算协议"。
- [第六章 · 封闭生态的采信逻辑:垂类权威与生态归属](https://canlah.ai/whitepaper/read/zh/09-chapter-6-how-closed-ecosystems-decide-what-to-trust-vert/) — 初稿中「多平台同质铺量可提升引用权重 35–50%」的主张,经独立复核后已被本报告撤回;本章记录复核过程与封闭生态真实的采信逻辑。
- [第七章 · 执行纪律:九十天推进序列与公开的方法论红线](https://canlah.ai/whitepaper/read/zh/10-chapter-7-execution-discipline-a-ninety-day-sequence-and-/) — 大多数 GEO 方案的执行顺序是"先做站内改造,因为它最容易交付"。
- [第八章 · 从这里开始](https://canlah.ai/whitepaper/read/zh/11-chapter-8-starting-from-here/) — 本报告通篇在说这个品类的承诺不可轻信。那么我们自己的承诺,也应该用同一把尺子量。以下是我们免费给什么、留联系方式能拿到什么、收费收什么,以及什么情况下我们会建议你别买。
- [附录 A · 方法与局限](https://canlah.ai/whitepaper/read/zh/12-appendix-a-method-and-limitations/) — 采集时间: 2026 年 8 月 11 日
- [附录 B · 不应引用的常见数据](https://canlah.ai/whitepaper/read/zh/13-appendix-b-commonly-circulated-claims-that-should-not-be-/) — 以下说法在本品类中流通广泛,经本报告独立复核后判定不可引用。

## Optional

- [Privacy Policy](https://canlah.ai/privacy/): How we handle data.
- [Terms of Service](https://canlah.ai/terms/): Engagement terms.
- [中文站点](https://canlah.ai/zh/): Simplified Chinese version of the site.