Reach developers and buyers from one site, with the evidence in one place.
Kanso analyses your public site, checks how AI crawlers can read your documentation and product pages, drafts technical articles from your Search Console data, and can turn a supported SEO fix into a pull request you review.
Free to start · No credit card required. See how it works
At a glance
Two audiences, many products and one stretched team
A technology company speaks to engineers who read documentation and buyers who read pricing pages, often across several products, with a small team.
Developers and buyers read different pages
Documentation, reference pages and pricing each answer a different reader, and each can fall behind the product.
Documentation is read by machines too
AI crawlers read docs and reference pages. A robots.txt rule or a missing llms.txt can shut them out without anyone choosing it.
Small fixes wait for engineering
A missing meta description or canonical link on a static page is a one-line change, and it queues behind product work.
From separate checks to one view of the site
Kanso reads the public site as a whole. It does not manage several products as separate brands.
- Auditing documentation, reference and pricing pages in different tools
- Finding out about blocked AI crawlers by chance
- Filing a ticket for each small metadata fix
- One analysis of up to 25 public pages, with evidence for each finding
- A readable status for each named AI crawler, llms.txt and structured data
- Supported metadata fixes prepared as pull requests for your own review
Analysis, readability, content and fixes
Three agents carry the core work. Coding applies where your site fits.
SEO Agent
Analyses up to 25 public pages and combines the result with Search Console data into recommendations.
- It produces
- Findings and tasks across your documentation and product pages.
- Where it fits
- The base layer every other step reads from.
GEO Agent
Checks nine named AI crawlers against your robots.txt, looks for llms.txt, and estimates how citable your pages look.
- It produces
- A readiness analysis and a labelled estimate.
- Where it fits
- Checks how machines read your docs.
Writer Agent
Picks a Search Console query where you are not on page one and drafts one long-form article in your brand voice.
- It produces
- A Markdown draft for review.
- Where it fits
- Starts technical content from real demand.
Coding Agent
For a supported finding on a static HTML page in your GitHub repository, prepares a small change and, once the owner confirms, opens a pull request.
- It produces
- A diff preview and a pull request.
- Where it fits
- Gets small metadata fixes into review.
From the whole site to a reviewed change
Six steps. The Coding step applies only when your pages are static HTML in a GitHub repository.
- 01You
Confirm the site that carries your products
A workspace works from one website, confirmed once and then locked. Choose the site your documentation and product pages live on, and connect Search Console, read-only.
Analyse the site
Kanso samples up to 25 public pages for titles, descriptions, canonical links, headings, links, robots.txt and sitemaps, and runs Lighthouse when you ask.
Check machine readability
It reports whether each of nine named AI crawlers is disallowed, whether llms.txt exists, and how your structured data looks. A pass means the file exists, not that any AI system uses it.
Draft from search demand
From the last 30 days of queries it picks one where you are not on page one and writes an article. It takes no brief, so topics follow your data.
Prepare supported fixes
For a supported title, description or canonical finding in static HTML, Kanso locates the file and shows the diff. Only the workspace owner can confirm a pull request, twice, and Kanso then stops.
- 06You
Review and merge yourself
Your engineers review and merge on GitHub under your own rules, and you publish drafts in your own tools.
What you receive
Each artefact appears before anything is created in your repository.
Site findings
Recommendations naming the page, the reasoning and the data, which can become tracked tasks.
A readiness analysis
A status for crawler access, llms.txt and structured data, with a model estimate labelled as one.
A technical article draft
Long-form Markdown for a query you already appear for.
A pull request
One commit on a new branch, labelled as a proposal, for supported static HTML changes.
Engineering stays in charge of the repository
Kanso works through GitHub on the one repository the owner chooses, and stops at a pull request.
Kanso does
- Reads your public pages and your search data
- Prepares a small, bounded change and shows the diff
- Opens a pull request on a new branch after confirmation
You decide
- Whether the workspace owner confirms a pull request
- Whether to merge it, under your own review rules
- What to publish and where
Kanso never
- Merges, approves or deploys anything
- Edits framework, script, configuration or workflow files
- Publishes a draft
Where this fits
Situations, not results. They depend on your site and your repository.
A static documentation site
If a page lacks a meta description, the Coding Agent can prepare the change as a pull request for your team to review.
Docs that should be machine-readable
The GEO Agent shows whether AI crawlers are allowed in and whether llms.txt is served.
A topic people search for
If Search Console shows searches for it on page two, the Writer Agent can draft a guide for an engineer to correct.
How Kanso approaches technical sites
Each point is something the product does today.
A pull request is a boundary your team trusts
Kanso's work ends at a reviewable diff, with your own review rules intact.
Changes are small and bounded
One file, at most 12 changed lines, and never a workflow, manifest or deployment file.
Machine readability has its own check
Crawler rules, llms.txt and structured data are reported apart from classic SEO findings.
What Kanso does not do for tech companies
A technology company often needs more than this. These are the limits.
It works from one website per workspace
A workspace is locked to one website, and the pricing page lists multi-brand workspaces as coming soon.
It does not edit framework sources
The Coding Agent changes one static HTML file, and stops when it cannot prove which file produces a page.
It does not merge or deploy
A person merges on GitHub, and your own pipeline deploys.
It does not produce documentation
It does not write reference pages, and it reads no code beyond the one HTML file in a pull request.
It does not publish drafts
There is no CMS integration. The citation view is a model estimate, and live AI answers are not measured.
Related use cases
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.
The GEO checks in more detail.
Turn Search Console Data Into Content
Turn Search Console opportunities into long-form article drafts in your brand voice, for you to review and publish.
The Writer workflow in more detail.
Ship Technical SEO Fixes
Turn supported technical SEO findings into GitHub pull requests you review. Kanso never merges or deploys.
The pull-request workflow in more detail, including its five supported changes.
SaaS
Search, content and community research for software teams whose marketing loses to product work.
For software teams whose product pages and search data are the focus.
Questions about this use case
Can Kanso handle several products?
On one website, yes: it analyses the site as a whole. A workspace is locked to a single website, and multi-brand workspaces are listed as coming soon on the pricing page.
Which GitHub changes can it make?
Five, in one static HTML file: add or fill a title, add or fill a meta description, and add a canonical link. The owner confirms twice, and you merge.
What if our site is built with a framework?
Kanso looks for the static HTML file and stops if it cannot prove which file produces the page. It does not edit framework sources.
Does Kanso check our documentation for AI crawlers?
It checks your public pages, your robots.txt rules for nine named crawlers and whether llms.txt is served. It does not measure whether an AI system uses any of it.
Can it write our technical guides?
It drafts one article per run from your Search Console data and takes no brief. You edit and publish it.
Does it need access to our code?
Only if you use the Coding Agent, and then only the one repository the owner chooses: its file list and one HTML file.
See plans and pricing · All use cases
See your whole site as developers, buyers and crawlers read it.
Enter your website. Kanso analyses it, checks how machines read it and lists what is worth fixing.
Get started free · No credit card required