More people now ask an assistant "who can build me a React Native app for a business in Australia?" than type it into Google. This post is about what those assistants actually read, and what I changed on this site so they can read it. It is technical, but the decisions are simple.
What an AI assistant reads
When ChatGPT, Claude or Perplexity answers a question with web results, a crawler fetched some pages and the model read the text. Three things follow from that:
- It reads HTML, not your JavaScript app. Most AI crawlers do not run JavaScript. A site that renders in the browser looks empty to them. This site is server-rendered; every word is in the HTML.
- It likes direct answers. A question followed by a two-sentence answer is easy to quote. A 1,500-word page with the answer buried in paragraph nine is not.
- It cross-checks facts. If my site says one thing about experience and the company site says another, the model trusts neither. Consistency across properties matters more than any single page.
What this site does about it
Allow the crawlers by name. The robots.txt explicitly allows GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot, Google-Extended and the others. Blocking them is the default on some hosts.
Serve an llms.txt. /llms.txt is a short Markdown summary: who I am, where, what services, links to every important page. /llms-full.txt is the whole site in one text file, including every FAQ and case study. An assistant that fetches either one gets the facts without parsing a layout.
Put the answer first. Every hire page, country page and service page opens with a direct answer to the question in its title, before any design or detail. The FAQs are real headings with short answers and FAQ structured data.
Describe the entities. The JSON-LD on every page says who the Person is, which Organization he founded, when, where, the phone number, the services, and links to the same profiles (LinkedIn, GitHub, YouTube) the company site links to. That is how a model learns that "Dilawar Khan" and "DevKey Technologies" are the same story.
Keep facts identical everywhere. Founded 2021, Islamabad, 50+ projects, the same phone number. On this site, on devkeytech.com, on LinkedIn. Where the company site was out of date, we updated it.
Do not duplicate the company site. devkeytech.com has the city pages and SME guides; this site has the hiring, outsourcing and country pages. Two sites competing for the same question would split the signal and lose to a third party.
What does not work
Keyword stuffing, pages that only swap a city name, and claiming things the rest of the web contradicts. Models are better at spotting those than search engines were ten years ago, because they are reading, not counting.
If you want this done for your business
AI-search readiness is part of the SEO service. The checklist above is the same one we run on client sites: server rendering, crawler access, llms.txt, answer-first pages, entity data, consistent facts. None of it is expensive; most of it is a week of work.