Recast's MCP Server Puts Live Device Data in AI Assistants' Hands
Recast unveiled an MCP Server for Right Click Tools at Microsoft Ignite, feeding AI assistants live endpoint data across its 60-million-endpoint install base while preserving existing permissions.

Why it matters
- Recast announced its MCP Server for Right Click Tools at Microsoft Ignite on November 17.
- The MCP Server retrieves live device data instead of relying on stale inventory records.
- Recast says its products manage more than 60 million endpoints.
- The integration requires tool-run approval and honors existing permissions, routing and licensing.
- Recast provides no independent benchmarks for investigation speed or diagnostic accuracy.
Recast announced an MCP Server for Right Click Tools at Microsoft Ignite on November 17, connecting natural-language AI assistants directly to live information from managed devices. The integration targets a bottleneck every IT team knows: gathering enough current context to understand a problem before deciding how to fix it.
The stakes are real. Recast says its products manage more than 60 million endpoints, extending Microsoft Intune and Configuration Manager across application delivery, automation, remediation and operational visibility. If conversational AI can speed up routine investigations at that scale, even modest time savings add up fast.
What does the MCP Server actually do?
Recast's MCP Server feeds AI assistants live device data instead of stale inventory records. That distinction sits at the center of the company's pitch. A record of a machine's earlier state may be useful for reporting, but troubleshooting often depends on what is happening at the moment an administrator investigates. Recast says its server retrieves information directly from devices, giving the assistant current context for diagnosis and for evaluating possible next steps.
The server acts as the bridge between an AI platform and the capabilities available through Right Click Tools. The model supplies the conversational interface; the connected software supplies the environment-specific information. An assistant can explain a common endpoint problem using general knowledge, but it cannot know the present condition of a particular corporate device without an authorized source of evidence. An MCP connection supplies that evidence without turning the model itself into the endpoint-management system.
Model Context Protocol, or MCP, is the standardized layer that makes this possible. It gives AI assistants a common way to connect with external tools and data. In this deployment, the protocol handles the plumbing between the assistant and the workflows Recast customers already run.
In short: the AI gets real evidence from real machines, not just a description pasted into a chat window.
How does it work with existing tools?
Recast positions the integration as another way to work with its existing tools, not a replacement for endpoint-management software. Administrators can ask questions about device and application issues, and the assistant draws on Right Click Tools workflows and data from their own environment.
The company also describes the integration as a way to combine multiple Right Click Tools workflows. That could reduce the manual work of moving between individual checks and assembling the results into a coherent picture. Checking whether a device is reachable, then inspecting a security-related setting, involves more than answering a general technical question — the assistant needs a connection to the right tools, access to the intended machine and permission to retrieve the requested information.
Recast's setup documentation describes using the server with:
- GitHub Copilot in Visual Studio Code
- GitHub Copilot CLI
- Device reachability checks
- BitLocker status retrieval
The documentation also shows tool-run approval as part of the workflow. A natural-language interface does not have to mean unrestricted execution; a human signs off before tools run against devices.
The takeaway: the assistant can chain checks and pull live results, but only with explicit approval inside permissions admins already control.
What happens to permissions and controls?
Recast emphasizes continuity with the controls administrators already use. The MCP Server honors existing permissions, routing and licensing rather than introducing an unrestricted path into the environment. For organizations evaluating the integration, that makes access design and tool scope as important as the quality of the assistant's answers.
"AI is most valuable when it gives IT professionals the context to make the right decision," said Will Teevan, Recast's CEO, in the announcement. The company's stated aim is to reduce manual effort while keeping administrators' expertise central to troubleshooting.
That framing matters for enterprise buyers. Many AI deployments raise questions about who owns a decision when a model suggests an action. Recast's answer is structural: the server respects the environment's existing guardrails, and admins stay in the loop.
One sentence: the AI never gets a side door — it inherits the same permissions and licensing already in place.
What's missing from the announcement?
Recast offers no independent benchmarks proving faster investigations or better diagnostic accuracy. The release does not provide measurements of how much investigation time the server saves, nor does it establish a measured improvement in diagnostic outcomes.
Teams evaluating the integration should distinguish two different claims. Faster information gathering is plausible when an assistant can pull live device data on request. Proven remediation outcomes are a stronger claim, particularly when a suggested action could affect multiple devices. The integration by itself does not establish that an AI-generated diagnosis is correct or that a recommended intervention should run without review.
Administrators still need to understand which devices were queried, which tools ran and what permissions governed the result. Live data improves the evidence available to an assistant; it does not remove the need for human judgment about what that evidence means.
The gap is simple to state: the announcement claims efficiency, and buyers will have to verify it themselves.
Why does this matter for enterprise AI?
The MCP announcement brings an AI-facing connection to Recast's existing management context rather than asking enterprises to replace their operational stack. That's a notable choice. Instead of pitching a chatbot as a new endpoint-management platform, Recast is testing a narrower role: helping IT professionals collect and interpret the information needed to act, while leaving operational accountability with the people responsible for the environment.
The practical question is whether conversational access makes routine investigations clearer and quicker without obscuring how a conclusion was reached. MCP-style integrations are becoming the standard pattern for connecting assistants to enterprise data, and endpoint management is one of the more obvious use cases — high volume, repetitive diagnostics, and a strong need for current state.
Recast's announcement points toward a more testable role for enterprise AI. The company has defined the job narrowly enough to measure: does live context help admins reach decisions faster, with the same controls in place? Organizations running the integration will find out whether that promise holds in practice.
Source: unite.ai
More from Sophie Lindqvist
Show full bio
Staff writer covering regulation, policy and their impact on artificial intelligence and machine learning at AI In Context.
749 articles
Related articles
- SyncHub says plain-English AI queries are replacing dashboards
- RHEL 10.3 and 9.9 Ship AI Diagnostics as Supported Production Feature
- Tanium launches AgentSight to track AI agents on corporate devices
- New Relic's Autopilot Kubernetes Agent Hits Public Preview
- Reco and Cyera integrate AI agent access risk ranking