Project deliveryHeadless CMS
When to rebuild your website platform (and when to resist)
Not every struggling platform needs a rebuild. Here's how to diagnose whether yours does, what the alternatives are, and how to make the ...
The Blog
We wrote an llms.txt for our agency site, here's what we put in it
Headless CMSTechnical SEO
Google is still the dominant way people find service businesses. But a growing number of prospects are asking AI tools questions that used to go to search. These tools don't crawl like Google, they need clear, structured information they can cite. We shipped an llms.txt file to make Little Dash legible to AI crawlers.
The shift
Search is splitting into two channels
Google is still the dominant way people find service businesses. That hasn't changed. But a growing number of prospects are asking AI tools, ChatGPT, Perplexity, Claude, questions like "who builds PayloadCMS sites in Australia" or "what does a headless CMS build cost."
These tools don't crawl the web the way Google does. They don't care about your meta description or your keyword density. They're looking for clear, structured information they can cite and summarise accurately. And most business websites give them nothing useful to work with.
This is what people mean when they talk about AEO (Answer Engine Optimisation) or GEO (Generative Engine Optimisation). It's not a replacement for SEO, it's a parallel channel that's growing fast.
We decided to do something concrete about it.
The file
What llms.txt actually is
`llms.txt` is a plain text file that sits at the root of your website, like `robots.txt`, but for AI crawlers instead of search engine bots.
The idea is simple: give language models a structured, machine-readable summary of your business that they can use to understand and cite you accurately. Not a sales pitch. Not marketing copy. A factual description of who you are, what you do, where you're based, who you work with, and where to find more detail.
It's not an official standard yet, it's an emerging convention. There's no guarantee every AI tool reads it today. But the cost of adding it is near zero, and the upside is that when these tools do look for structured business information, yours is there and accurate.
Think of it as a clarity investment, not a ranking hack.
The structure
What we included in ours
We structured our llms.txt around the information an AI tool would need to accurately describe Little Dash in a recommendation or citation.
- 01Business summary
Who we are, where we're based, what we specialise in, and who we work with. One paragraph that gives an AI enough context to describe us without hallucinating details. This is entity clarity, making sure the machine knows Little Dash is a Brisbane-based web development studio, not a fitness brand or a SaaS product.
- 02Service pillars with context
Each service area listed with its URL, a brief description, and indicative pricing and timeline information. Not a full sales page, just enough for an AI to answer "how much does a PayloadCMS site cost" or "how long does a headless build take" with something grounded in reality.
- 03Work and case studies
Links to completed projects with short descriptions. These give AI tools concrete examples to reference when someone asks "show me agencies that have built on PayloadCMS" or similar.
- 04Company pages and FAQ
Links to about, testimonials, FAQ, and contact pages. The FAQ page is particularly important it's structured content that directly answers the kinds of questions AI tools are asked. We also reference our FAQ content architecture to make the connection between content modelling and AI readability.
- 05 Social and verification links
LinkedIn, Instagram, Facebook URLs. These help AI tools verify that Little Dash is a real business with an active presence, not a placeholder or SEO farm.
The pairing
llms.txt doesn't work alone
The file itself is just a pointer. It's only useful if the pages it links to are clear, well-structured, and genuinely informative.
This is where the connection to content architecture matters. Our FAQ page is modelled as grouped, topic-based accordions with anchor IDs — not a single dumped list. Our service pages describe what we actually deliver, with enough specificity for an AI to extract meaningful answers. Our headless CMS pillar page covers the topic comprehensively enough to be cited as a source.
llms.txt tells AI tools where to look. The quality of what they find when they get there determines whether you get cited accurately, cited badly, or ignored entirely.
The honesty
What we don't know yet
AEO is not like traditional SEO. There's no equivalent of Google Search Console for AI citations. You can't reliably track when an AI tool recommends your business, how often it happens, or what triggered it.
The standards are still emerging. `llms.txt` might become widely adopted, or it might be superseded by something else. The AI tools themselves are changing how they source and cite information every few months.
What we do know is that clear, structured, factual content about your business is never wasted. Whether it's consumed by Google, by an AI tool, or by a human reading your site directly, the investment in entity clarity pays off across every channel.
We're treating llms.txt as one piece of a broader content architecture strategy, not a magic bullet, but a practical step that costs almost nothing to implement and positions the site for wherever AI-powered search goes next.