AI Visibility Audits·Consumer AI Agents·Anthropic Claude · OpenAI·ChatGPT · Perplexity · Claude · Google AI Overviews·llms.txt · schema.org · /api/profile · /api/chat·klava.london/ai·AI Visibility Audits·Consumer AI Agents·Anthropic Claude · OpenAI·ChatGPT · Perplexity · Claude · Google AI Overviews·llms.txt · schema.org · /api/profile · /api/chat·klava.london/ai·
AI Visibility · Consumer AI Agents

AI visibility
& consumer AI agents.

Klava London delivers two AI service lines end-to-end: AI-search visibility (making sure ChatGPT, Perplexity, Claude and Google AI Overviews can find, read, cite and recommend your site) and consumer AI agents (on-site concierges, booking bots, product finders and PDF-to-structured-data pipelines grounded in your actual business).

Two halves of the same shift. Outside the product: the website itself needs to be structured so LLMs can trust and cite it. Inside the product: AI agents replace the manual steps that slow teams down — bookings, support triage, data entry, information lookup.

A concrete example on the visibility side. I've run 20 AI-visibility audits so far (two paid engagements, optimisation delivered for selected clients) and built the free public tool at klava.london/audit: enter a URL, get a report back in ~15 seconds on how ChatGPT, Perplexity, Claude and Google AI Overviews see your site, with a copy-paste llms.txt you can drop in today. No email, no sign-up.

A concrete example on the agents side. At Global Chic Club — a 5,000-member events and B2B membership platform across 20+ countries — an admin used to retype every event by hand from PDF flyers. Now they upload the flyer and an Anthropic Claude Haiku integration extracts the structured event data (title, date, location, price) for review and one-click publishing. The manual step is gone. It runs daily. On the Dr Glyn Medispa clinic site, an on-site AI concierge answers treatment, pricing and booking questions from the actual clinic knowledge base.

Every site I build ships with the visibility layer by default: /llms.txt, full schema.org JSON-LD, a machine-readable /api/profile endpoint, semantic HTML and clean structure — so the LLM economy works for the businesses I build for, not around them.

If you want an AI visibility audit or a consumer AI agent scoped for your business, let's talk.

Pillar 01 — AI-search visibility
/llms.txt — a plain-text file at the root of your domain, written for AI models. Tells them who you are, what you do, and how to recommend you. Like robots.txt — but for LLMs.
schema.org — JSON-LD markup for Person, Organization, Service, FAQ and BreadcrumbList. Gives Google and AI crawlers unambiguous, machine-readable facts about your business — increasing citation confidence and rich-result eligibility.
/api/profile — a read-only machine-readable endpoint AI agents can query directly for contact, availability, services and rates. No scraping required.
Semantic HTML & content — clean heading hierarchy, descriptive anchor text, and content structured to answer real user questions directly. AI cites pages whose meaning is unambiguous.
robots.txt AI-crawler allowlist — explicit permission for GPTBot, ClaudeBot, PerplexityBot, OAI-SearchBot, Google-Extended, Applebot-Extended and Bingbot. Most sites still accidentally block them.
Pillar 02 — Consumer AI agents
On-site concierges — grounded in a business's actual knowledge base: treatment info, pricing, opening hours, booking flow. Answers what a receptionist would, with citations. Live example: Dr Glyn Medispa clinic site.
PDF-to-structured-data pipelines — vision-capable Claude models extract typed, validated JSON from admin-uploaded documents. Replaces manual data entry. Live example: Global Chic Club event flyers, running daily.
Booking & support automation — first-line triage, appointment questions, product finders, order lookups. Human-in-the-loop for anything that touches money or medical advice.
Anthropic Claude & OpenAI — production code, monitored cost, structured outputs, retries. Not demos, not prototypes — features that survive contact with real customers.
Live on this site
/llms.txtPlain-text — written for LLMs/api/chatLive Claude assistant/api/profileMachine-readable JSON/auditPublic AI Visibility Auditschema.org JSON-LDPerson · Organization · FAQ · Service
Run a free audit →See all services
In short

Two halves of the same shift. AI-search visibility: get the site cited by ChatGPT, Perplexity, Claude and Google AI Overviews. Consumer AI agents: on-site concierges and workflow automations grounded in your real knowledge base — not generic chatbots.

A process, simplified
Before
Admin retypes every event by hand from a PDF flyer — title, date, location, price.
Step 01
PDF flyer in — admin uploads the flyer through the dashboard.
Step 02
Claude Haiku — Anthropic's vision-capable model reads the flyer and extracts the event data.
Step 03
Structured JSON — title, date, location, price, ticket URL — typed and validated.
Step 04
Review & publish — admin confirms in one click. The manual step is gone.
Scale
5,000+ members across 20+ countries. Runs every day.
Why both halves
58%
of Google searches now return an AI Overview before organic results.
3×
more likely to be cited by AI search tools when a site has correct schema.org markup and an llms.txt file.
∞
hours saved when manual data entry is replaced by a Claude pipeline that runs in seconds.
Two AI service lines delivered end-to-end — visibility on the outside, agents on the inside — so the LLM economy works for the businesses I build for, not around them.
klava.london
London, UK