How people find professional services in 2026
In 2023, someone looking for a fractional commercial director in London would have typed a query into Google, scanned the first page of results, and clicked through to two or three websites. In 2026, a significant and growing proportion of those same searches happen through AI assistants.
The person opens ChatGPT, Perplexity, or Claude and types: "Who are the best fractional growth directors in London for a professional services firm with around thirty people?" The AI generates a response. It names businesses. It describes what they do. It explains why they might be a good fit.
The businesses it names are not chosen randomly. They are the ones whose websites the AI could read, understand and trust. Businesses whose sites were invisible to AI crawlers, or whose content was too vague or unstructured for the AI to extract a clear description from, are simply not in the answer.
This is the new visibility problem. And most SME websites have it.
What AI systems need to cite your business
An AI assistant deciding whether to recommend a business in response to a query is doing something like the following: it retrieves information from sources it has indexed or can access in real time, assesses whether each source is credible and relevant, and synthesises a response that it believes accurately answers the question.
For your business to appear in that response, three things need to be true. The AI needs to be able to access your site (many AI crawlers are blocked by default Cloudflare settings or absent robots.txt permissions). It needs to understand what your business does, who it serves and what makes it credible. And it needs to be confident that what it says about you is accurate, because AI systems are sensitive to the reputational cost of hallucinating incorrect information.
Most business websites fail on all three.
The four elements of AI search optimisation
1. llms.txt
A llms.txt file is a plain text file placed at the root of your website that tells AI language model crawlers what they are permitted to access and how the site is structured. It is the AI equivalent of robots.txt. Without it, many AI systems default to treating your site as restricted.
A well-structured llms.txt file lists the key pages of your site, describes what each contains, and explicitly signals that AI crawlers are welcome to index the content. It can also include a condensed version of your business description that AI systems can use when summarising who you are.
2. The ai-agent-summary meta tag
This is a meta tag in your page's HTML head section that gives AI crawlers a concise, accurate summary of the page's content. Unlike the standard meta description (which is written for human readers and search engine snippets), the ai-agent-summary is written specifically for AI systems, clear, factual, complete enough to be cited directly.
Every page on your site should have one. The homepage summary should describe the business as a whole. Service page summaries should describe the specific service. Article summaries should describe the argument and conclusions of the piece.
3. Structured data schema
Schema markup is code added to your HTML that gives search engines and AI systems structured, machine-readable information about your business. For a consultancy, relevant schema types include Organisation (your business name, address, contact details, areas of expertise), Person (the founder's credentials and professional history), Service (what each service involves and who it is for), and FAQPage (common questions and answers).
Schema is the difference between an AI system guessing what your business does from reading your homepage copy, and knowing it from structured data that has been explicitly provided for that purpose.
4. Clear, citable content
The most durable AI discoverability signal is content that is written clearly enough to be cited accurately. AI systems are trained to be cautious about recommending sources they cannot summarise reliably. Content that makes clear arguments, states specific facts and conclusions, and is structured with headings that signal the topic of each section is far easier for an AI to work with than content that is vague, marketing-heavy or structured primarily for aesthetics.
This does not mean dry or mechanical writing. It means writing with enough precision that an AI reading your service page comes away with a clear understanding of what you do, who you do it for, and why it works.
Why this matters more for professional services than almost any other sector
Professional services clients, those seeking consultants, advisers, solicitors, financial planners, specialists of any kind, are exactly the type of buyer who turns to AI assistants for research. They have a complex, high-stakes decision to make. They want a synthesis of the landscape before they commit to any outreach. AI assistants are well-suited to that synthesis.
The businesses that appear in those syntheses are the ones that have done the work. The businesses that do not appear are not losing visibility to a better competitor. They are losing it to no one, because the AI simply cannot find them.
The window for early adoption advantage is open now. It will not be open indefinitely.