AI bringing network support closer to the moment of Need #
When someone reports, “The Wi-Fi is slow,” the hardest part is often not fixing the problem. It is figuring out what the problem actually is.
The request may begin with a user, move to the helpdesk, get escalated to a network team, and then make its way through several tools before anyone has enough context to act. By that point, the user is still waiting, the helpdesk is chasing information, and a network engineer may be investigating a symptom instead of a cause.
AI can change that experience by bringing network context into the places where support work already happens.
With Cisco’s new Model Context Protocol (MCP) servers, organizations can connect network data to AI-powered assistants, ticketing workflows, portals, and custom applications. The goal is not to add another destination for IT teams to monitor. It is to make the network more useful wherever a support decision needs to be made.
From “Wi-Fi is slow” to a better ticket #
Consider a common helpdesk interaction. An employee opens a ticket saying that video calls are dropping in one part of the office. The L1 analyst does not need to become a wireless specialist before taking the first step.
An AI-enabled workflow, built into a service-management platform like ServiceNow, could gather relevant network context and help the analyst answer practical questions:
- Is the issue isolated to one user, device, access point, or location?
- Are other users experiencing the same problem?
- Is the access point healthy?
- Are there authentication, connectivity, or coverage signals that point to a likely cause?
- What information should be included before the ticket is escalated?
Instead of forwarding an incomplete report to a networking team, the helpdesk can attach a more useful first assessment. The network team receives better evidence, the user spends less time repeating the same story, and the organization can reserve specialist attention for the issues that genuinely require it.
The value is not that an AI system replaces the helpdesk or the network engineer. The value is that it helps each person start with more relevant context.
Self-service before the support case #
Why stop at the helpdesk? the same capability can move even closer to the end user.
Imagine an employee whose laptop is struggling to connect to wireless. Before opening a case, the employee asks a self-service assistant to check the connection. The assistant can help determine whether the problem appears related to the device, the user’s location, the access point, authentication, or a broader service condition.
Sometimes the answer may be simple: move closer to an access point (we didn’t plan coverage in the staircase), reconnect to the appropriate network, or retry after an authentication issue. In other cases, the assistant may confirm that the issue is worth reporting and create a case with useful diagnostic context already attached.
That creates a better experience on both sides of the support boundary. Users get an answer sooner, while support teams receive fewer vague tickets and more actionable information when escalation is necessary.
These experiences are examples of what organizations can build around network data. They are not claims that every organization will get the same workflow out of the box. MCP gives developers and IT teams a way to connect Cisco network capabilities to the applications and assistants they already use.
Choosing the right experience: AI Assistant or MCP? #
Cisco’s MCP servers are complementary to the AI experiences already available in Cisco products.
Cisco AI Assistant is the natural choice for people who are operating and troubleshooting their Meraki environment within the Meraki experience. It helps network teams ask questions, understand conditions, and act in that product context.
MCP is the better fit when an organization wants to bring network intelligence into a broader workflow. That might mean a helpdesk assistant, an employee self-service portal, a cross-vendor operations copilot, an automated report, or a custom agent that combines network information with data from other systems.
The distinction is straightforward:
Use AI Assistant when the work starts in the Cisco product. Use MCP when the work starts in the workflow your organization is building around the product.
This gives organizations more choice without forcing one experience to do every job.
For some organizations, however, the choice is not simply about preference. Security policy may require all LLM processing to use an organization’s own managed or private models. Other organizations may not be permitted to send operational data to a vendor-hosted AI experience or may operate networks that are completely air-gapped. A locally deployed, open-source MCP server gives these organizations a path to build AI-enabled workflows around approved models and infrastructure. For Catalyst Center environments, the MCP server and the AI experience can run within the same controlled network boundary. Self-hosting Meraki MCP provides deployment and customization control, but it does not make Meraki Dashboard itself air-gapped; live Meraki data still requires an approved connection to the cloud-managed Dashboard.
Hosted simplicity or open-source control #
Different organizations need different levels of control over where software runs and how it is adapted. That is why Cisco is making MCP available through both hosted and open-source approaches.
Hosted Meraki MCP: the lowest-friction path
The Cisco-hosted Meraki MCP server is designed for teams that want to connect an AI client to Meraki with minimal operational overhead. Cisco manages the service, while customers can focus on the workflows they want to create.
For many organizations, that simplicity is hard to beat. There is no server to package, operate, or update. Teams can start with a hosted service and explore use cases such as network-aware ticket triage, reporting, and support assistance without first building an integration platform.
Open-source Meraki MCP: adapt the experience to your environment
The open-source Meraki MCP server is for organizations that want to run the software in their own environment, inspect the implementation, and customize it to meet their operational or compliance requirements.
Developers can adapt the code to fit internal workflows, connect it to their preferred AI environment, or extend the experience around the needs of their organization. This is especially useful when the deployment model, network environment, or governance requirements make a Cisco-hosted service less suitable.
Open source also creates transparency. Teams can review how the integration works and make informed decisions about how it fits into their own architecture. Customized deployments remain the responsibility of the organization operating them, but the starting point is available to the community rather than hidden behind a proprietary integration.
Open-source Catalyst Center MCP: local deployment for controlled environments
For Catalyst Center customers, the open-source MCP server provides a locally deployable way to connect AI experiences to network inventory, device health, wireless experience, software, compliance, and other operational data. Local deployment matters for organizations with strict data, infrastructure, or sovereignty requirements. It can also support environments where the AI system and Catalyst Center need to remain inside a controlled or air-gapped network. In that model, the MCP server runs alongside the systems it needs to reach instead of depending on a public hosted service.
That control is often essential in regulated industries, government environments, and other organizations that need to audit source code, manage their own infrastructure, or customize the integration before placing it into production.
Open source is an invitation to build #
Making the code available is more than a licensing decision. It is an acknowledgement that our users and developers will imagine workflows we have not yet anticipated.
Some teams will use the servers as delivered. Others will build a service-desk experience, an internal operations portal, a private AI assistant, or a multi-vendor workflow around them. The open-source projects give those teams a foundation they can inspect, adapt, and operate according to their own requirements.
One possible output is a network-health view that summarizes the evidence, highlights the highest-impact conditions, and suggests where investigation should begin. The examples below are illustrative outputs from an AI application built with network data accessed through the Meraki MCP server. They are not interfaces generated by the MCP server out of the box.
The next support interaction can start with context #
Network support does not have to begin with a vague ticket and a series of handoffs. A helpdesk analyst can start with better evidence. An end user can get useful guidance before opening a case. A network engineer can spend more time solving the difficult problems instead of collecting basic facts.
Cisco’s MCP servers give organizations a way to build those experiences around the environments they already operate. Choose the hosted Meraki MCP server when simplicity and speed are the priority. Choose open source when local operation, auditability, and customization matter most. For Catalyst Center environments with strict deployment requirements, local open-source operation can bring AI-enabled workflows into places a hosted service cannot.
The larger opportunity is simple: make the network available at the moment a decision needs to be made – and make that decision easier to act on.
Choose the path that fits what you want to build:
Start with the Cisco-managed Meraki hosted MCP serverSelf-host or customize the open-source Meraki MCP serverDeploy the open-source Catalyst Center MCP server