The London AI Visibility Index — 25 Brands Ranked by AI Search Presence

We audited 25 London tech and AI companies for AI search visibility. The average score is 59 out of 100. Only one brand, DeepMind, scored above 88. Fourteen brands score below 60, putting them beyond the reach of ChatGPT, Gemini, and Perplexity when users research their product category. Here is the full ranking.

How we built the index#

BrandCited is an AI visibility intelligence platform that monitors AI platforms for brand citations and shows you exactly what to fix to appear more often in AI-generated answers. For this index, BrandCited applied a 6-signal technical audit to each of the 25 domains. The signals are: presence of an llms.txt file, AI crawler permissions in robots.txt, Organization schema markup on the homepage, homepage accessibility, a populated meta description, and homepage content depth above 300 words. Each signal contributes a fixed point value to the overall score, with a maximum of 100.

The 25 brands were selected from London's most active tech sectors: fintech, AI infrastructure, cybersecurity, and applied AI. Sources used to identify the companies: Sifted's UK AI coverage, Beauhurst's UK startup database, Crunchbase London AI search, Tech Nation's State of the Nation report, and G2's UK software vendor listings. Run a free full audit at brandcited.ai.

What the London AI Visibility Index found#

• Only 1 of 25 brands (4%) scored above 88 out of 100, the threshold BrandCited associates with consistent AI citations across ChatGPT, Gemini, and Perplexity.
• 96% of brands (24 of 25) are missing an `llms.txt` file, the AI-specific discoverability file that tells crawlers who you are, what you do, and which content carries the most authority.
• 60% of brands (15 of 25) have no Organization schema markup on their homepage, leaving AI models to infer entity identity from unstructured page text.
• The average score across all 25 London brands was 59 out of 100, compared to a Silicon Valley benchmark of 62 and a global cross-industry average of approximately 35. London's top brands perform well, but the median masks a long tail of companies that are near-invisible to AI search.
• The sharpest pattern: London's AI companies score no better than fintech or SaaS brands. Tractable builds AI for insurance claims and scores 55. Featurespace builds adaptive AI for fraud detection and scores 50. Eigen Technologies builds document AI and scores 45. Each of these companies builds AI that depends on structured, machine-readable data. Their own sites lack the structured data that would make them readable by AI search engines.

The full London AI Visibility Index ranking#

Brands marked "est." received a quick 6-signal audit. Run a free full audit at brandcited.ai to see your complete score across 30+ signals.

What the top London brands are doing right#

DeepMind scores 90 out of 100, the highest in this index. As Google's AI research lab and one of the most cited organisations in machine learning, DeepMind has built web infrastructure that reflects its technical culture. Their robots.txt allows GPTBot, ClaudeBot, PerplexityBot, and Google-Extended, with minor path restrictions on certain documentation sections that cost them 10 points from a perfect score. Organization schema is present on the homepage with sameAs references to Google Scholar profiles, their Wikipedia entry, and verified social accounts. Their homepage carries over 300 words of dense, factual prose about research areas, products, and mission. DeepMind is also the only brand in this index with an llms.txt file at their domain root, a signal that directly tells AI crawlers how to understand and cite the organisation.

Revolut scores 80 out of 100. The London fintech, valued at over £33 billion in its 2024 funding round and serving 45 million customers across 35 countries, passes five of the six key visibility signals. Organization schema is present on the homepage. All major AI crawlers have full access. Homepage content exceeds 300 words of structured, factual prose about products, markets, and company positioning. Meta descriptions are populated. The one missing signal is llms.txt. For a company that processes over 400 million transactions per month and employs engineers who build payment infrastructure from the ground up, deploying a 50-line text file is the most fixable gap in this entire index.

Wise scores 80 out of 100 and carries the same signal profile as Revolut: five passing, one missing. Wise's homepage is structured for extraction in a way that few London brands match. The first 200 words answer who uses Wise, what problem it solves, how it differs from banks, and what it costs. That is the exact information AI models need to form a citation. Organization schema on their site links to their London Stock Exchange listing, their Crunchbase profile, and primary social accounts, giving AI models a cross-referenced entity declaration across public data sources. Adding llms.txt would close the only gap between Wise and the top tier.

The most common AI visibility gaps in London#

llms.txt is a plain-text file at your domain root that tells AI crawlers who you are, what you do, and which pages carry the most authority. Only DeepMind has one in this index. The other 24 brands leave AI models to construct their entity understanding from unstructured page content, a process that produces inconsistent results across ChatGPT, Gemini, and Perplexity.

Here is a minimal template you can deploy at yourdomain.com/llms.txt in under 30 minutes:

# llms.txt for [Company Name]

## About
[Company Name] is a [one-sentence description: what you are, what you do, who you serve].

## Key pages
- Homepage: https://[yourdomain.com]/ — overview of the product
- Product: https://[yourdomain.com]/product — feature details
- Pricing: https://[yourdomain.com]/pricing — plan and pricing
- Blog: https://[yourdomain.com]/blog — research and guides

## What we do
[Two or three sentences describing your product. Direct, factual language. No marketing copy. AI models use this text when answering questions about your company.]

BrandCited checks for llms.txt in the first pass of every audit and provides a generated template based on your homepage content.

Organization schema is a JSON-LD declaration on your homepage that tells AI models your brand name, URL, description, founding date, and social profiles in structured format. Without it, a model encounters "Eigen Technologies" or "Quantexa" in crawled content and builds an inference rather than reading a declaration. Inferences can be wrong, incomplete, or confused with similarly-named companies.

Add this block to your site layout's <head> section:

{
  "@context": "https://schema.org",
  "@type": "Organization",
  "name": "Your Company Name",
  "url": "https://yourdomain.com",
  "description": "One sentence: what you are and what you do.",
  "foundingDate": "2018",
  "sameAs": [
    "https://linkedin.com/company/your-company",
    "https://twitter.com/yourhandle",
    "https://crunchbase.com/organization/your-company"
  ]
}

The sameAs array is the most underused field in Organization schema. It cross-references your entity across public data sources and gives AI models a verified entity declaration they can anchor citations to. BrandCited's data shows Organization schema is the single highest-impact fix for brands scoring below 65.

Six brands in this index have homepages with fewer than 300 words of visible prose: Signal AI, Quantexa, Eigen Technologies, Behavox, Tessian, and Freetrade. AI models extract entity descriptions from page text. A homepage built around imagery, animations, and short feature callouts gives a language model almost nothing to work with.

The fix is not a long homepage. It is a homepage that answers five questions in plain prose:

1. 1What is this company?
2. 2What does it do?
3. 3Who does it serve?
4. 4What problem does it solve?
5. 5What makes it different?

Three hundred words of clear answers to these five questions pass the content depth signal. Testimonial carousels and feature icon grids do not.

How does your brand rank?#

We audited 25 London tech brands. If yours is not on the list, or received an estimated score, run a free full audit at brandcited.ai. BrandCited checks 30+ signals across five categories and shows your complete score in 30 seconds, with every issue ranked by impact.

The brands that scored well in this index share one trait: they treat their website as a machine-readable entity declaration. DeepMind, Revolut, and Wise all built sites that answer the questions AI models need before they will cite you.

Run a free AI visibility audit at brandcited.ai. Your score across 9 AI platforms in 30 seconds, every issue ranked by impact.

Frequently asked questions#

The London AI Visibility Index is a BrandCited research report ranking 25 London-based tech and AI companies by their AI search visibility score. BrandCited is an AI visibility intelligence platform that monitors AI platforms for brand citations. The index checks six technical signals per domain and produces a score from 0 to 100. BrandCited publishes city-level AI visibility reports to help brands benchmark against their closest market peers.

Building AI products does not mean your website is optimised for AI search engines. Tractable builds AI for insurance claims and scores 55. Featurespace builds adaptive AI for fraud detection and scores 50. Eigen Technologies builds document AI and scores 45. The gap is not technical capability. It is a failure to apply AI-readability principles to their own marketing infrastructure. BrandCited shows exactly what to fix and ranks each issue by expected score impact.

London's average of 59 sits just below Silicon Valley's 62 from BrandCited's April 2026 index. Both markets far exceed the global cross-industry average of approximately 35. The key structural difference: only one London brand — DeepMind — has deployed an llms.txt file, compared to four brands in Silicon Valley. BrandCited publishes city-level AI visibility reports monthly.

llms.txt is a plain-text file placed at your domain root that tells AI crawlers who your company is, what you do, and which pages carry authority. Only DeepMind has one in this index. The file is a relatively new standard, formalised in late 2024, and adoption has been slow outside AI-native companies. BrandCited checks for llms.txt in the first scan of every audit and generates a tailored template based on your site content.

Add Organization schema to your homepage layout, publish an llms.txt file at your domain root, and verify that GPTBot and PerplexityBot are not blocked in your robots.txt. These three changes address the three most common gaps found in this index. Run a BrandCited audit after implementing to confirm the changes are detected and identify remaining gaps. Most brands see measurable improvement in AI citation rates within four to six weeks of fixing these signals.

Manual testing — asking ChatGPT "what is [company]?" and noting the response — gives a point-in-time check on one platform. BrandCited tests your brand across 9 AI platforms on a weekly schedule, tracks citation rates over time, and benchmarks results against London competitors in the same sector. Manual testing misses platform-specific citation patterns and takes hours per session. BrandCited surfaces those patterns across all platforms and tells you which signal to fix first.