Help AI systems find, understand and cite your site.
Kanso checks the signals AI systems can read on your website, estimates how citable your pages look, and turns what it finds into evidence-backed actions you can track. It does not ask ChatGPT, Gemini or Perplexity about you, and it says so.
Free to start · No credit card required · See how it works
Being found now includes being understood by AI
More questions go to AI systems, and those systems can only use what they can reach, read and make sense of.
AI systems need sources they can read
When someone asks an AI a question, the answer draws on pages the system can reach and make sense of.
Machine-readable signals are easy to overlook
Files and markup such as llms.txt and structured data are not visible to visitors, so they are rarely checked.
Crawl rules can quietly block access
A robots.txt written for another purpose can end up disallowing AI crawlers without anyone intending it.
Clear content is easier to use
An answer that is buried, vague or split across pages is harder for a system to quote.
Evidence supports trust
Claims that are specific and backed up are easier to rely on than general statements.
Nobody sees it in one place
The checks are spread across files, tools and opinions, with no single list of what to fix.
Without Kanso, and with it
- Unsure whether AI crawlers can reach your site
- No clear picture of your llms.txt or robots.txt state
- No single audit of the signals AI systems read
- Technical findings that are hard to prioritize
- No evidence behind the advice you get
- No tracked list of what to do next
- A technical readiness analysis of your public site
- Each AI crawler checked against your robots.txt rules
- A model-based estimate of how citable your pages look, labelled as an estimate
- Recommendations that show the evidence and the limits of that evidence
- Action tasks you can track to completion
- The analysis run again to see whether signals now pass
SEO helps search engines understand your site. GEO helps AI systems use it.
Generative engine optimization is a young field with no single formal definition. This is how Kanso uses the term.
Discoverability
Can a system find your pages at all: a sitemap, a crawlable site and canonical links.
Machine accessibility
Do your rules allow AI crawlers, and does the site serve files such as llms.txt.
Content clarity
Are titles, headings and visible text clear enough to be understood and quoted.
Evidence
Is what you say specific and supported, with structured data that states who you are.
Citation potential
How likely a page looks to be useful as a source for a question. Kanso estimates this with a model.
Continuous improvement
Signals change as your site does, so the checks run again and show what moved.
What Kanso observes, what it estimates, and what it does not measure
GEO tools are easy to oversell. Kanso keeps these three apart, in the product as well as on this page.
Readiness signals
- llms.txt present and readable
- Each of nine AI crawlers allowed or disallowed in robots.txt
- Whether the site can be crawled
- Structured data, titles, descriptions, headings and canonicals
- Sitemap and HTTPS
Deterministic checks of your public site. No model is involved.
Citation potential
- Whether a model thinks an AI assistant would cite your site for a few inferred queries
- A one-sentence reason for each
- A tracked competitor judged more likely to be cited
- A model-estimated score
Kanso's own model judgment. It is not a result from any AI product.
Live AI answers
- What ChatGPT, Gemini, Perplexity or Claude actually say about you
- Citations or rankings over time
- Share of answers
- Whether any AI system uses your llms.txt
Kanso does not query AI platforms, and says so in the product.
From your public site to a tracked action in five steps
Inspect
Kanso reads what your public website exposes to AI systems.
- Samples up to 25 public pages, plus your robots.txt, sitemaps and llms.txt
- Checks each of nine named AI crawlers against your robots.txt rules
- Looks for structured data, titles, descriptions, headings, canonical links and visible text
- Reports every signal as pass, warning, missing or not measured
Evaluate
The GEO Agent estimates how citable your pages look, and labels that as an estimate.
- A model judges whether an AI assistant would cite your site for a few queries inferred from your site
- Each judgment comes with a one-sentence reason
- When you track competitors, it names the one it judges more likely to be cited
- Runs when you run the GEO Agent, and a daily run is built in but currently paused. It is Kanso's estimate, not a result from any AI product
Recommend, with evidence
Findings become recommendations that carry the facts they rest on.
- Written from facts Kanso collected, which it resolves itself rather than taking from the model
- Observed data is shown apart from what the data cannot establish
- Wording is labelled as an AI-generated suggestion, never as fact
- You can read, question and dismiss each one
Act
A recommendation can become work that is tracked.
- A task moves from open to in progress to marked complete by a person
- Every change is recorded with who made it
- Kanso states what it does not know, such as which file or system produces a page
- Kanso does not edit your website or open pull requests for GEO changes
Re-check
Run the analysis again to see whether the signals moved.
- Each readiness signal updates, so you can see whether llms.txt, crawler access or structured data now passes
- A passing signal means the file or markup is there, not that any AI system uses it
- Kanso's formal verification covers page-level SEO conditions today and does not yet cover GEO tasks
- So Kanso never calls a GEO task verified
What the GEO Agent does
GEO readiness analysis
Checks the public signals AI systems can read, across a sample of up to 25 pages, weekly or when you run it.
AI crawler checks
Reports whether each of nine named AI crawlers is disallowed from your homepage in robots.txt.
llms.txt check
Looks for an llms.txt file, tells you whether it is readable, and flags it when it is missing.
Structured data and page signals
Checks Organization and WebSite markup, structured-data coverage, titles, descriptions, headings, canonicals and visible text.
Crawl accessibility
Confirms the homepage can be fetched, is not noindex and is not blocked by robots.txt, and looks for a valid sitemap.
Citation-potential evaluation
A model-based estimate, per inferred query, of whether an AI assistant would cite you, with a reason. Runs when you run the GEO Agent. Scheduled runs are built in and currently paused.
Evidence-backed recommendations
Each lists the observed data behind it and what that data cannot establish. The wording is labelled as an AI-generated suggestion.
Action tasks
Turn a recommendation into a task that moves from open to in progress to marked complete by a person, with an audit trail.
Answers that show their sources
When you ask the Kanso CMO about GEO, its answer separates what was observed, derived, interpreted by AI and not measured.
Evidence, then action, then a second look
Many GEO tools stop at a score. Kanso keeps the evidence next to each recommendation and carries the work forward.
Every recommendation shows its evidence
The facts behind a recommendation are resolved by Kanso from its own data and marked as observed, derived or a limitation.
Estimates stay labelled
The citation-potential view is called what it is, a model-based estimate, and the product states that live AI visibility is not measured.
Changes stay with people
Kanso records tasks and who did what. It does not change your site, and it does not claim a task is verified when it cannot check it.
Related, not identical
Both depend on a site that can be crawled and understood. They ask different questions of it.
| SEO | GEO | |
|---|---|---|
| Aim | Be found and understood by search engines | Be found, understood and potentially cited by AI systems |
| Starting point | Keywords and search intent | Questions, context and clear answers |
| Technical signals | Titles, canonicals, sitemaps, speed and links | AI crawler rules, llms.txt, structured data and readable text |
| Crawlers | Search crawlers | Search crawlers and AI crawlers |
| Where results appear | Search results | AI-generated answers, which Kanso does not measure |
| In Kanso | The SEO Agent | The GEO Agent |
GEO complements SEO and does not replace it.
Kanso has both. See the SEO Agent.
Kanso GEO Agent, manual research, or a generic AI assistant
| Manual research | Generic AI assistant | Kanso GEO Agent | |
|---|---|---|---|
| Project-specific evidence | You check your files by hand | Only what you paste into the prompt | A public analysis of your own site |
| AI crawler accessibility | Read robots.txt yourself | Cannot inspect your robots.txt | Each of nine named crawlers checked against your rules |
| llms.txt and structured data | Checked manually, if at all | Cannot check your live site | Checked as readiness signals |
| Citation potential | Your own judgment | One model's opinion of what you paste | A model-based estimate for queries inferred from your site, labelled as an estimate |
| Evidence-backed recommendations | Written up by whoever audits | Free-form text | Typed, with observed data and stated limits |
| Action workflow | A separate task tracker | None | Tracked action tasks |
| Live AI engine results | Ask each product yourself | One product's answer, not a measurement | Not provided. Kanso does not query AI platforms |
Tools and workflows vary. This describes the typical case of working by hand, or with a general AI assistant that is not connected to your site.
Part of Kanso, not a separate tool
The GEO Agent works from the same company documents as Kanso's other agents. Start with your website, and see the pricing page for which plans include it.
Got questions?
GEO Agent questions
Where the GEO Agent is used
These use cases name this agent in their workflow. Each one says what it produces and where you stay in control.
Get Found in AI Search
Check how readable your site is to AI systems, estimate how citable it looks, and turn the gaps into actions.
SaaS
Search, content and community research for software teams whose marketing loses to product work.
Tech Companies
Search and content support for technology companies that sell to developers and buyers.
Ecommerce
Search and content drafts for online stores, built from your own site and Search Console data.
B2C
Social and content drafts for consumer brands that need to stay consistent across channels.
Law Firms
Practice-area content and search drafts for law firms. Kanso does not check compliance or give legal advice.
Know how your site reads to AI systems.
Kanso checks your AI-facing signals, finds what is worth fixing and helps you track the work. Enter your website to start.
Get started free · No credit card required