Leave the conversation to find the context
- Open the KNWN Visibility dashboard.
- Select the correct site and report.
- Find the relevant gap, question, page, or strategy.
- Carry that context back into the working conversation.
KNWN VISIBILITY MCP CASE STUDY
We made KNWN Visibility usable inside AI assistants.
KNWN Visibility is our AI visibility platform. We built a focused, read-only MCP server so users can ask for visibility summaries, gaps, questions, recommendations, and strategies without leaving their AI workflow.
7 focused tools
Live product data
Read-only launch
Ask KNWN Visibility from an AI client
Powered by MCP
01
“Where are we losing visibility?”
The user asks a product question in ChatGPT or another compatible client.
02
get_visibility_gaps
The MCP chooses the focused product tool and checks access to the requested site.
03
Live product answer
KNWN Visibility returns the permitted gaps and context from the product API.
Product permissions remain in control
No direct database access
THE PRODUCT OPPORTUNITY
The dashboard had the answer. The conversation had the question.
KNWN Visibility already tracks brand visibility, prompts, citations, gaps, recommendations, and saved strategies. The dashboard is where users explore the full product, but it is not where every question begins.
A marketer planning content in ChatGPT might want a quick visibility summary. A founder might want to see where competitors appear instead. A content lead might need the next questions or page recommendations without stopping the conversation to search through several screens.
We wanted those answers to come from KNWN Visibility itself. The MCP makes selected product workflows available inside an AI client while the dashboard, API, permissions, and business rules remain the source of truth.
The MCP is not a second version of KNWN Visibility. It is a new way to use the product.
WORKING EVIDENCE
The source product, the live tool surface, and the observability layer.
BEFORE AND AFTER
Less context switching. Faster access to the same product intelligence.
The MCP does not replace the dashboard. It removes unnecessary navigation when the user already knows the question they need answered.
Ask KNWN Visibility directly
- Ask a normal product question inside the AI client.
- Let the MCP choose the focused tool.
- Verify the user's access to the requested site or page.
- Return the permitted live product data in the conversation.
WHAT USERS CAN ASK
Normal questions in. Focused product tools behind the scenes.
The user does not need to know a tool name. They ask about visibility, gaps, questions, pages, or strategy and receive data from the KNWN Visibility account they are allowed to access.
How visible are we right now?
Returns the selected site's current visibility summary and recent direction from KNWN Visibility.
get_visibility_overview
Where are we losing visibility?
Surfaces the gaps that matter for a site the user is already authorized to access.
get_visibility_gaps
Which questions should we cover next?
Finds relevant questions that can become new content or product-positioning opportunities.
get_trending_questions
What should we improve on this page?
Returns page recommendations after confirming that the page belongs to the selected site.
get_page_recommendations
THE TOOL SURFACE
Seven tools built around real KNWN Visibility jobs.
We deliberately did not mirror the full API. The launch surface covers a small set of useful questions while keeping risky or inconsistent operations out.
| Tool | What it does |
|---|---|
list_sites | Show the KNWN Visibility sites the signed-in user can access. |
get_visibility_overview | Return a site's current visibility summary and recent direction. |
get_visibility_gaps | Find weak or missing visibility opportunities for an authorized site. |
get_trending_questions | Surface questions that create a relevant visibility opportunity. |
get_page_recommendations | Return recommendations for a page that belongs to the selected site. |
get_strategies | Read the user's saved KNWN Visibility action plans and their status. |
get_usage_snapshot | Show how the MCP is being used and where requests encounter friction. |
HOW IT WORKS
The product remains in control.
KNWN Visibility remains responsible for identity, tenant access, object ownership, and product data. The MCP handles the AI-facing tool contract and request flow.
One account connection
The account-linking flow keeps the AI-client token separate from the KNWN Visibility product credential. Product access stays server-side.
Permission-aware requests
Every site and page request is checked against the signed-in user's existing KNWN Visibility access before data is returned.
A read-only launch surface
The MCP can retrieve insights and recommendations, but it cannot publish, spend credits, enqueue jobs, or change account data.
01
The user connects their KNWN Visibility account to a compatible AI client.
02
The client calls one focused MCP tool with the requested site or page.
03
The MCP verifies scope and confirms that the user can access the requested object.
04
The KNWN Visibility API returns only the permitted data needed for the answer.
Current availability: the public MCP demo and seven-tool server are live. Customer account linking remains in controlled rollout while we complete the production OAuth configuration.
OBSERVABILITY
We instrumented the MCP from day one.
The analytics layer helps us see which tools people use, where requests slow down, and whether a technically successful call actually resolved the task.
What the MCP records
- Tool name, timing, outcome, and result count.
- The authorized KNWN Visibility site involved in the request.
- Observable friction such as missing input, partial results, rate limits, or repeated calls.
- Tool completion separately from full request resolution.
What it deliberately excludes
- Authorization headers, access tokens, cookies, and API keys.
- Full upstream responses, raw email addresses, and IP addresses.
- The complete ChatGPT or Claude conversation.
- Hidden model reasoning or chain of thought.
WHAT WE LEARNED
The protocol was the easy part. Product boundaries were the real work.
Building the MCP against a product we operate made the trade-offs concrete. The same lessons apply to any SaaS product that wants to become useful inside AI clients.
Start with questions, not endpoints
The useful tool surface was not a copy of the KNWN Visibility API. It was a small set of product questions users already need answered.
Keep the product as the source of truth
Identity, account access, site ownership, page ownership, and business rules remain inside KNWN Visibility rather than being recreated in the MCP.
Launch read-only before adding actions
A narrow read surface is easier to test, explain, monitor, and trust before introducing any operation that changes data or spends credits.
Measure usefulness, not just calls
Our analytics separates a technically successful tool call from an answer that actually resolved the user's request.
FAQ
Questions about the KNWN Visibility MCP server.
What is KNWN Visibility?
KNWN Visibility is the AI visibility platform built and operated by KNWN. It tracks how brands appear in AI answers, the prompts and citations shaping that visibility, competitive gaps, and recommended actions.
What can a user do through the MCP server?
A connected user can list their KNWN Visibility sites, retrieve a visibility overview, inspect gaps, find trending questions, request page recommendations, read saved strategies, and review MCP usage.
Does the MCP replace the KNWN Visibility dashboard?
No. The dashboard remains the complete product experience. The MCP is a faster conversational route into selected product workflows when a user is already working inside ChatGPT or another compatible AI client.
Can the MCP change customer data?
No. The current tool surface is intentionally read-only. It does not publish content, spend credits, start background jobs, or modify KNWN Visibility records.
How does the MCP enforce account permissions?
The MCP uses KNWN Visibility's existing account and object permissions. Every site or page request is checked before product data is returned, and the connector does not access the product database directly.
Can KNWN build this kind of MCP for another SaaS product?
Yes. We design the tool surface, authentication, tenant checks, write boundaries, analytics, deployment, and handover around the product's real workflows and existing permission model.
BRING YOUR PRODUCT INTO AI WORKFLOWS
Let customers use your product without leaving the conversation.
Show us the product, the account model, and the questions or actions that matter. We will design the smallest MCP surface that makes those workflows useful, secure, and operable.