10 things we're doing to get our website found by AI

Twice this week I asked a new lead "How'd you hear about Kowalah?"

"I asked AI" was the response.

Here are a few things we're doing that are helping us get found and referenced by AI.

1. April Dunford positioning

April wrote 'the' book on positioning — Obviously Awesome — which helps you describe what 'shelf' your potential customer is looking at and where they are most likely to find you in their digital search.

Run her positioning exercise first and get clear on who you serve, how you are different, and what value you provide to them.

2. Corey Haines skills

I'm going to mention a few of Corey's skills - they come in a plugin called "Marketing Skills" and they are gold dust for whoever runs your website.

Your first step having installed the plugin in Claude Code or Codex is to run the product context skill which takes your positioning, and any other background, to create a markdown file in your website project that the other skills will refer to.

3. Astro.build web framework

Not to be confused with the new ChatGPT model of the same name, Astro is a web build framework (think where you might have used Wordpress or Wix previously). It is designed for this AI driven content world, coming with an MCP server that helps your AI agent understand how to use it and build.

4. Claude Code

You could pick Codex or your AI of choice, the key is that we build all the web pages using AI, not marketing developers hand rolling code. It means we can spin up new pages, styles and sections for the site quickly to respond to what we are hearing from clients and prospects.

Use Corey's copy writing and copy editing skills to ensure what you write supports your positioning.

5. Granola

We use Granola to transcribe all our prospect and client calls. This is exceptionally useful for our consultants, but also for our marketing team who can surface relevant topics and content areas for us to add to our webinar and blog series

6. LLMS.txt

The jury is out on whether adding an llms.txt file to your website route helps your site get found by AI models. Our view is it can't hurt. Our llms.txt guides AI models to the key parts of our site, including our past webinar topics, our AI cost calculator and our services offerings.

7. SEO audit/AI-SEO Audit - Corey Haines skills

This skill does a review of your current site, the product marketing context you provided, plus any other context you provide (Granola) to derive new opportunities to get found by search engines and AI models

8. WebMCP - AI Cost Calculator

This is a new proposed standard from Microsoft and Google engineers. It allows AI tools to interact with your website via tools you expose. We have an interactive page that allows companies to plan their AI programme budget and we expose three tools to allow an AI model to drive that process. Only available via ChatGPT desktop today we expect to see visits using agents and MebMCP increase

9. Integrate web into other channels

Referrals are our biggest channel - customer A experiences Kowalah, refers wife/husband/ex-colleague. We have a referral programme to recognise, reward and encourage this and use it as a final slide in every workshop and training — kowalah.com/refer — this drives traffic to the site that might otherwise have not existed.

10. Lead magnets

This final one is a combination of all of the above - we create un-gated lead magnets "Organise your AI week", "Cowork set up plugins", "AI hackathon facilitator guides" and these sit on the site, answering the top questions our target ICP is facing. While these may be a downloadable PDF, we ensure they are written directly as content on the site, easily accessible and scrapable by AI models, ready to be cited in responses.

Use Corey's Lead Magnet skill to develop ideas that will support your ideal customers where they are in their buying journey.

These are just a few of the things we are doing - if I was to summarise the overall approach it isreducing the gap from idea to live to near zero.

We don't want any idea to be stuck in a monday.com table ready for possible review in Q4.

Idea > Build It > Publish > Learn > Iterate