More AI citations a day across 102 agent-ready pages, from 123 to 219 in four weeks, as AI engines started citing the company on MCP and agent security questions.
COMPANY OVERVIEW
The company, at a glance.
A venture-backed AI security company that helps enterprises discover, test and protect AI agents, MCP servers and AI coding assistants. It used Fibr AXO so the AI engines answering security buyers’ questions could understand its website and quote it.
INDUSTRY
AI and agent security
BUYERS
AppSec leads, security engineers, CISOs
MARKET
US enterprise
USE CASE
AI search visibility with Fibr AXO
The program at a glance
102
Agent-ready pages
4
AI engines measured
50
Buyer prompts tracked daily
4 weeks
From first launch to results
Security buyers come with four kinds of questions: learning how a new AI threat works, building a vendor shortlist, working out what it takes to deploy a control, and asking about the company by name or against a rival. Each sits at a different point in the buying journey, and an AI engine is now as likely to answer it as a page of Google results.
In a Nutshell
AI engines found the company by name. On MCP and agent security questions, it was often left off the list.
Security buyers increasingly ask ChatGPT, Gemini, Claude or Google’s AI Overviews which AI and MCP security vendors to evaluate. When they asked about the company by name, the engines found it. When they asked about the threats it protects against, it was often missing. With Fibr AXO, the company made 102 pages agent-ready in two launches. Four weeks later, those pages get 219 AI citations a day, up from 123 in launch week, and weekly prompt coverage rose from 24 to 40 of 50 tracked buyer questions.
THE CHALLENGE
AI engines found the company by name, but missed it on MCP and security questions.
Content wasn’t the problem. Product pages, threat guides, deployment docs and comparison articles already covered every topic buyers asked about. But navigation, diagrams, scripts and claims spread across sections made pages hard for an AI engine to read, judge and quote with confidence.
The cost of being invisible
When the company did appear, engines cited 942 third-party pages alongside it, more than four times its own. Analyst lists and rival blogs were explaining the category to buyers, and every unbranded question it missed was a shortlist it wasn’t on.
THE SOLUTION
One website, two audiences: humans and AI agents.
The company ran Fibr AXO in three steps: making the gap visible, mapping the questions security buyers actually ask, and shipping agent-ready pages in two launches, measured every day.
STEP 1
The Analysis: making the gap visible
Daily tracking showed two clear problems. 8 of 44 core pages got no AI citation at all in launch week, and the deeper MCP, guardrail and red-teaming pages, the ones that answer specific buyer questions, were often missing. Each engine also read the site very differently: ChatGPT cited the company’s pages 807 times in launch week, Claude just 78.
STEP 2
The Signals: four intents and four engines
AXO was built around the questions buyers ask an AI engine, not the keywords they type into Google:
Informational: buyers asking how MCP, guardrail and red-teaming risks work. 6 prompts.
Commercial: buyers asking which AI and MCP security vendors to evaluate. 24 prompts.
Transactional: buyers asking what it takes to roll out a control. 5 prompts.
Branded: buyers asking about the company by name, or against a named rival. 15 prompts.
Every intent was tracked daily on every engine, because each engine finds and cites pages in its own way.
STEP 3
The Setup: 102 agent-ready pages in two launches
Launch 1 (14 to 15 Oct 2026) covered 44 core pages: the homepage, agentic security, MCP security, AI guardrails, AI red teaming, the vendor shortlist guide and alternatives pages. Launch 2 (25 Oct 2026) added 58 long-tail pages: MCP threat guides, AI coding assistant guides, deployment docs, integrations, research reports, blog guides and FAQs. Every page and prompt was tracked daily with Promptwatch until 8 Nov 2026, with no changes to what human visitors see.
The New Agent Experience
A human visitor sees the same company website as before. An AI engine gets an agent-ready version of the same page: what the product is, who it’s for, how it compares and how it deploys, in a form it can read and quote. The goal isn’t just to be found. It’s to be understood well enough to be cited as evidence.
THE RESULTS
Launch week (12 to 18 Oct) compared with the latest full week (2 to 8 Nov 2026), tracked daily with Promptwatch across ChatGPT, Google AI Overviews, Gemini and Claude. Illustrative results, modelled on Fibr’s own AXO program.
DAILY AXO PAGE CITATIONS
↑ 78%
From 123 to 219 a day across the 102 AXO pages. The 58 long-tail pages added about 44 of those.
PROMPTS THAT BRING IT UP
↑ 67%
From 24 to 40 of 50 a week. 46 of 50 at least once since launch.
ALL SITE CITATIONS PER DAY
↑ 62%
From 201 to 326 a day across every page on the company’s site.
AXO PAGES CITED
82 of 102
Cited at least once since launch, up from 36 in launch week.
The gains came from unbranded questions: informational prompts that bring up the company went from 2 to 6, commercial from 6 to 21, and transactional from 1 to 4.
ONGOING RESULTS
The first launch did most of the work. The homepage stayed flat at about 27 citations a day, while the other 43 first-launch pages went from 96 to 148 a day (+54%). The 58 second-launch pages added about 44 citations a day in the latest week, and 40 of them were cited within their first two weeks.
Growth is slowing: the latest week was up 9% on the week before, after 29% the week before that. Claude is still the slowest engine to pick up the new pages.
KEY INSIGHTS
INSIGHT 01
Your brand name is not an AI search strategy.
All 15 branded prompts found the company before AXO, and every gain came from unbranded questions. Buyers asking before they have a shortlist often couldn’t find it, and that is where AI engines now decide who gets considered.
INSIGHT 02
Depth beats the front door in AI search.
The homepage stayed flat while the 43 pages behind it rose 54%. Engines cite the page that answers the specific question, so pages on one product, one use case or one comparison are where AI visibility is won.
INSIGHT 03
Every engine reads your site differently.
Prompt coverage grew 120% on Google AI Overviews and 19% on Claude. One overall “AI visibility score” would hide that gap. Tracking each engine separately shows where the work pays off, and where it doesn’t yet.
