MCP server for your website · SEO 3.0
Found by AI. Understood by AI.
An MCP server for your website: a digital catalogue of your company, products and services that AI agents and people use through the same interface. VISEON Ask is MCP for people.
VISEON for Web delivers SEO 3.0, Semantic Entity Optimisation. We build and govern the knowledge graph of your organisation, its brands, products, services and people, and run the graph as an MCP (Model Context Protocol) server on your domain. AI assistants query it over MCP; people use the same MCP interface through VISEON Ask. Both draw on the same facts, with their sources. Then we list your MCP server in the official MCP Registry and publish an ARD manifest, so agents find you and understand you. Then your customers’ agents can transact, B2B and B2C, on facts they can trust.
Part of SEO 3.0: a granular, entity-by-entity assessment of how AI sees your organisation today, so the results are accurate, not a generic score.
- MCP server on your domain
- Listed in the official MCP Registry
- ARD registry verified
- Schema.org & JSON-LD
- Ask is MCP for people
1,461
entities in our own governed graph, live on viseon.io.
6
MCP tools, independently verified by an ARD registry.
0
logins needed to read the VISEON knowledge graph. Public by design, read-only.
<10 ms
for exact graph answers, served from the edge.
Why now
Search is now AI. AI answers from what it understands.
Google has turned search into AI-enabled search, and assistants such as ChatGPT, Claude, Gemini and Perplexity answer questions about your business directly. AI assistants don’t rank your pages; AI assistants assemble an answer from the entities they can identify and trust.
If your organisation, products and services are not declared as clear, connected entities, AI fills the gap from third-party content: reviews, directories, old articles and competitors who described themselves better. That is digital obscurity, and ranking well on Google does not cure it.
Illustration: the sources an AI assembles an answer from
SEO 3.0: optimise the entity, not the page
Semantic Entity Optimisation gives each thing your organisation is and offers one canonical definition, a stable identifier, verified links to the rest of the web and a record of where each fact came from. AI then reads you from the source, not from what others say about you. SEO 3.0 still includes everything schema markup has always had to do, Google Rich Results included; SEO 3.0 just doesn’t stop there.
How it works
One MCP hub. Three spokes.
Every interaction with your organisation, by a person or by AI, goes through one MCP interface: Interact, on your own domain. People use the Interact MCP interface through VISEON Ask, agents connect to the VISEON MCP server directly, and browser agents use WebMCP. From there, three spokes cover the whole journey, from being found to being bought from.
InteractEvery interaction on your domain, by people and by AIAsk is MCP for people · MCP for agents · WebMCP in the browserWhat’s included
Everything SEO 3.0 takes, in one service
From the markup you already have to a graph that agents can find, query and act on.
Validate
Schema.org and JSON-LD audited across your domain, Google Rich Results checked, and the output of Yoast, Rank Math and other plugins included.
Govern
One canonical entity each for your organisation, brands, products, services and people, with a stable identifier, sameAs links and provenance.
Serve
The graph served from your own domain and from the edge, with exact answers in under 10 milliseconds: one MCP interface, with VISEON Ask for people, the MCP endpoint for agents, and WebMCP where needed.
Publish
An Agentic Resource Discovery (ARD) manifest on your domain and a listing in the official MCP Registry, verified as yours.
Built on open standards: Schema.org, JSON-LD, the Model Context Protocol, NLWeb and ARD. The graph is portable, inspectable and yours.
The hub
One MCP interface. For people and for agents.
VISEON Ask is MCP for people: visitors ask in plain English and get answers from your governed graph, with links to the pages behind them, instead of a list of search results. The same MCP interface serves AI assistants directly, and WebMCP serves agents acting in the browser. Because people and agents use one interface over one source, VISEON Ask and AI agents work from the same facts, and context shapes how each answer is put.
- What does this company actually do?
- Which plan covers five domains?
- Who leads sales?
- Is this product in stock?
- Exact answers first. A direct graph answer where one exists, evidence retrieval where the graph has none, and a language model only to phrase what the evidence supports.
- Every answer shows its source. Responses link to the pages behind them, and say plainly when the graph does not know.
- Live where it matters. Constants, such as who you are and what you offer, come from Schema. Transients, such as price and stock, are streamed in real time through change data capture (CDC), so answers stay live.
- Answers from the edge. The graph is served at the edge, so exact graph answers return in under 10 milliseconds, to people and agents alike.
- Public by design. Only what you choose to declare is served. VISEON for Web is read-only, needs no login, and exposes nothing private.
- Open to any assistant. Any MCP-capable assistant can query your catalogue, whichever model it runs on.
Proof, in public
We run VISEON for Web first. On this site.
VISEON.IO is built exactly the way we build yours: every page, product, service, person, glossary term and FAQ is an entity in one governed graph, served from viseon.io to people and to agents.
- VISEON Ask answers from the graph at viseon.io/ask.
- An MCP endpoint at
viseon.io/mcp, supporting both current revisions of the MCP specification. - An ARD manifest at
/.well-known/ard.json. - Listed in the official MCP Registry as
io.viseon/semantic-intelligence, verified as ours through our domain. - All MCP tools independently verified by an ARD registry, Neuronto.
1,461
entities in the viseon.io graph.
6
MCP tools, independently verified.
2
discovery listings: ARD and the MCP Registry.
0
logins needed to read the viseon.io knowledge graph. Public by design.
Where VISEON for Web fits
Keep your CMS and your plugins. Add what they can’t do.
Schema plugins generate markup page by page. Schema platforms manage it, and some serve it to agents from their own infrastructure. VISEON for Web audits and governs whatever markup you have, then serves the result from your domain through VISEON Ask for people, an MCP endpoint for agents, and ARD and MCP Registry discovery listings.
Nothing is ripped out. Yoast, Rank Math or your platform’s schema keep working; VISEON makes the whole domain one coherent graph.
VISEON for Web: complete on its own
Your organisation, brands, products, services and people, declared for AI search, assistants and agentic commerce. VISEON for Web needs nothing else to work.
Also available: VISEON for Enterprise
An optional, separate product for your internal estate: the same hub and technology over operations, policies and rules, served to your own agents behind your permissions.
Getting there
From audit to published, in four steps
Start with the free Semantic Entity Assessment, a granular, entity-by-entity part of SEO 3.0, see exactly where you stand, and build from there.
Audit
Minutes
Your free Semantic Entity Assessment: your domain’s graph assessed entity by entity for missing, invalid and conflicting entities, Rich Results and reachability, so the findings are accurate and specific.
Build
One to three DAYS
On your domain, included in your plan: we agree the change with you, configure sync and a validation baseline, then validate and test.
Connect
Weeks
VISEON Ask on your site and an MCP endpoint on your domain, with WebMCP where agents need to act.
Canonical entities and identifiers, shaped to your sector. We draft your ontology straight away; the timeline is set by how quickly your team agrees it, typically one to three months.
Publish
Then ongoing
ARD manifest and MCP Registry listing, then monitoring as your site and offer change.
Governance, weekly catalogue health reports, change monitoring, and live prices and stock through change data capture.
Plans
Start free. Scale by brand.
Semantic Entity Assessment
£0
one complimentary report
- Domain-wide health scores
- Entity-level findings, not a generic score
- AI discoverability gaps
- Schema accuracy and completeness
Single Brand
from £7,000
per year, three-year commitment
- One brand or domain, up to 10,000 pages
- Weekly catalogue health reports
- VISEON Ask and your MCP endpoint
- ARD manifest and MCP Registry listing
- Live prices and stock through CDC*
Multi-Brand
from £28,000
per year, three-year commitment
- Up to five domains and 100,000 SKUs
- Everything in Single Brand, including live CDC*
- A schema architecture designed for your brands
- Direct integration with your own AI and LLM pipelines
- Dedicated account manager
*Live prices and stock need a change data capture tool, licensed separately. WebMCP, for agents acting in the browser, is part of our paid Transact services.
Larger estates and full plan details: see pricing.
Questions
What people ask first
What is SEO 3.0?
SEO 3.0 is Semantic Entity Optimisation: optimising the entities your organisation is and offers, rather than individual pages or keywords. Each entity gets one canonical definition, a stable identifier, verified links and provenance, so AI search and AI agents can identify, understand and cite each entity directly.
Does VISEON help with Google Rich Results?
Yes. Rich Results are part of SEO 3.0: VISEON audits check that your Schema.org markup meets Google’s requirements for enhanced results such as FAQ snippets, product cards, review stars, events, breadcrumbs and organisation panels. The difference is that we don’t stop at the markup: the same entities are governed as one graph and served to AI directly.
How is SEO 3.0 different from GEO?
Generative engine optimisation tries to influence what AI engines write about you, mostly through content. It rarely holds, because those engines assemble answers by retrieval-augmented generation (RAG), which is probabilistic, so the same question can get a different answer each time, and costly, because every answer is retrieved and generated afresh. SEO 3.0 gives AI an authoritative source to read instead: your own governed graph, served from your domain, with exact answers wherever the graph has them.
What does an MCP endpoint on our website do?
An MCP endpoint on your website lets AI assistants and agents query your organisation’s knowledge graph directly, over the Model Context Protocol, instead of scraping and guessing. The MCP endpoint is the direct agent connection to Interact, the hub on your domain; VISEON Ask is the same MCP interface for people, answering your visitors from the same graph. Where agents need to act in the browser, WebMCP extends the MCP endpoint.
What are ARD and the MCP Registry, and why publish to them?
Agentic Resource Discovery (ARD) is an open specification for a manifest on your domain that tells agents and registries what you offer. The official MCP Registry is the public directory of MCP servers that other directories draw on. Publishing to both, verified through your domain, is how agents find your MCP endpoint without being told about it.
Will VISEON work with WordPress SEO plugins?
Yes, and nothing is replaced. VISEON audits and builds on the schema that Yoast, Rank Math and other plugins produce, resolves conflicts and gaps across the whole domain, and adds what plugins do not provide: one governed graph, VISEON Ask, an MCP endpoint and discovery listings.
How is this different from AI-visibility monitoring tools?
Monitoring tools report what AI assistants say about you today. Monitoring tools do not change what those assistants can read. VISEON for Web publishes the governed, canonical catalogue that assistants read from, on your own domain, and the free assessment shows where your catalogue is missing or in conflict. The two work together: monitor the output, govern the source.
Can’t we build this ourselves with NLWeb or other open standards?
NLWeb, Schema.org, JSON-LD and MCP are the building blocks, and VISEON is built on them, which is why the result is portable and yours. What the service adds is the work around them: auditing and reconciling the schema you already have, one canonical entity for each organisation, product, service and person, answers that cite their source and say when the graph does not know, discovery listings (ARD and the MCP Registry), and governance as your site and offer change. Standing up an endpoint is the easy part; keeping the catalogue correct is the ongoing job.
Be understood by AI, not just found
Start with a free semantic entity assessment, part of SEO 3.0: a granular, entity-by-entity view of how clearly AI can see your organisation, products and services today, and what would change it.
Or email [email protected] · Built by Differentia Consulting, enterprise data specialists since 2002.