llms.txt on HubSpot: How to Prepare Your Site for the Age of AI Assistants

Stop and think about how you discovered the last tool or service you hired. More and more, the answer is not "I searched Google and clicked a link". It is "I asked an AI assistant and it recommended one". That shift in discovery behavior is happening now, and it brings a new discipline alongside the usual SEO.

That discipline is AEO, answer engine optimization. And one of the pieces that goes into this strategy is llms.txt, a file that helps language models understand your site. In this guide, I will explain what llms.txt is, why it matters now, and how to publish it on HubSpot Content Hub, so your content has a better chance of being understood and cited by AI assistants.

What llms.txt is

llms.txt is a text file, placed at the root of the site, that presents the site's content cleanly and in a structured way for language models. Instead of the model having to dig through HTML full of menus, scripts, styles, and navigation elements to understand what the site is about, it finds a direct summary: who you are, which pages matter, and what each one covers.

The idea is a relative of robots.txt, that file that guides search crawlers, but aimed at artificial intelligence instead of traditional search engines. It is a way to speak the models' language, delivering your content in the format they process best, instead of leaving them to guess from a page full of noise. It is a gesture of technical courtesy that increases your chance of being understood well.

Why this matters now

When an AI assistant answers a question about your industry, it decides what to cite based on what it can understand and trust. A site that is hard to interpret loses ground to one that explains itself well and in a structured way. This is not about tricking the model, it is about making its life easier so your content, which is good, is effectively recognized as a source.

The crucial point is that AEO does not replace SEO, it adds to it. The same quality content needs to be readable for three audiences at once: the people who read, the search engines that rank, and the AI models that summarize and cite. Whoever ignores that third front is left out of exactly the discovery channel that is growing the most, ceding ground to competitors who prepared.

How to publish llms.txt on HubSpot Content Hub

On Content Hub, the reliable way to publish llms.txt is to upload the file to the file manager, leave it public, and then set up a proxy redirect from /llms.txt to that hosted file. The proxy is the key part: unlike a 301, which would send visitors to the file's own address, it serves the file's content while keeping /llms.txt as the address that answers. Point the redirect at the file's URL on your connected domain, not the default HubSpot one, so the file goes out with the correct plain-text content type.

That keeps llms.txt clean at the HTTP level, delivered as a raw text file with no page wrapper and no tracking scripts loading on the URL. To update it later, you replace the file in the file manager and the redirect keeps pointing to the same place, so /llms.txt reflects the new version. When new reference content is added, or the page architecture changes, refreshing the file is a quick, contained task, not a technical project outside the platform.

What to put in the file

A good llms.txt is direct and useful, with no empty marketing. It describes the site in a way a model quickly understands who you are and what you offer of most value. The essential elements are few and clear. It follows a defined Markdown structure, not free-form text: an H1 with the site or company name, which is the only required part, a short summary in a blockquote right below it, and then one or more H2 sections listing the pages as Markdown links, each link followed by a brief note. When you can, point those links to clean Markdown versions of the pages, the same URL with .md appended, since that is the format the models read best.

• A line of identity: who the company is and what it does, objectively.
• The core pages, like products, services, and resources, each with a short description of what it covers.
• The reference content you want cited, like guides and in-depth technical articles.
• Clean, factual text, focused on what actually describes your site, with no promotional flourish.

The golden rule is clarity. The more direct and well-organized the file, the easier it is for the model to extract and use the information. Think of it as an annotated index of your site, written for whoever needs to understand the whole in seconds, not to impress. Usefulness beats rhetoric in this format.

AEO and SEO go together

It is worth insisting on this point because it is misunderstood: investing in AEO does not mean abandoning SEO, it means broadening the strategy. The same best practices that help content rank, clarity, heading structure, direct answers, real depth, also help an AI model understand and cite it. Well-made content serves both worlds.

llms.txt is an extra layer that complements that well-made content, giving the models a clean map of your site. Together, solid SEO and good AI preparation cover the whole spectrum of how people discover content today, from the traditional search engine to the conversational assistant. Preparing for both is what keeps your brand visible regardless of where your audience chooses to look.

Why this matters for your operation

AI discovery is already a reality and tends to grow. Being present and well-understood in that channel is what ensures your brand keeps being found as the audience's behavior changes. Whoever prepares early reaps the advantage of already being well-positioned when the channel matures, instead of chasing after it once competitors have taken the space.

Ignoring that front, on the other hand, is betting that discovery behavior will stand still, which will not happen. A company that invests only in traditional SEO and does not prepare for the age of assistants risks seeing its traffic and visibility migrate to a channel where it does not appear. Preparing the site for AI is protecting the brand's discovery in the long run.

In practice: the good content AI did not cite

A company had excellent technical content on its blog, but noticed that, when asking AI assistants about the topics it dominated, it was competitors who showed up cited. The content was there, it was good, but it reached the models buried in heavy HTML, hard to interpret, with no map pointing to what was most relevant.

Part of the answer was preparing the site for AI, including publishing an llms.txt that presented the company and pointed to the reference content cleanly, together with strengthening the answer-first structure in the articles. Over time, the content came to be more recognized as a source. The lesson was that having good content is not enough; you have to deliver it in a way the models can understand and trust.

Checklist of AEO preparation

1. Is there an llms.txt at the domain root, served as plain text by Content Hub?
2. Does the file carry the company's identity and the core pages with short descriptions?
3. Is the reference content you want cited listed?
4. Is the text clean and factual, with no promotional flourish?
5. Is the content pointed to actually good and structured, not shallow pages?
6. Is llms.txt updated when the site's structure or content changes?

Frequently asked questions

Does llms.txt replace traditional SEO?

No. It complements it. SEO handles discovery by search engines, and llms.txt helps AI assistants understand and cite your site. Good, well-structured content serves both at once.

Where does the file need to live?

At the domain root, on the standard /llms.txt path, served as plain text. On HubSpot Content Hub, you get this by uploading the file to the file manager as public and adding a proxy redirect from /llms.txt to it.

Do I need to update llms.txt often?

Update it when the site's structure changes or when new reference content is added. Because it lives as a file behind a proxy redirect, updating means replacing the file in the file manager, and the redirect keeps pointing to it.

What is AEO?

AEO is answer engine optimization, the practice of preparing content to appear and be cited in AI-powered assistants. Unlike SEO, which focuses on ranking, AEO focuses on being understood and cited by the models.

Does llms.txt guarantee AI will cite me?

It does not guarantee it. It increases the chance by making it easier for models to understand your site, but the foundation remains quality, well-structured content. It amplifies good content, it does not create visibility from nothing.

Is AEO worth it if my traffic comes from Google?

Yes, because discovery behavior is changing and part of searches already pass through AI answers. Preparing early protects the brand's visibility as the audience migrates to assistants, without abandoning SEO.

Want your site understood and cited by both Google and AI assistants? At Insight Sales we prepare your HubSpot for SEO and AEO in an integrated way. Talk to our team and find out how we can help.