What Are Customer Engagement Platforms with MCP Support?
When customer engagement platforms advertise MCP support, it means the platform can be navigated and controlled via MCP using natural language. An MCP server, also known as Model Context Protocol, connects to AI tools such as Claude, ChatGPT, or Cursor and allows users to perform read and write actions via a pre-built secure connection.
A customer engagement platform is software that helps brands reach, serve, and reward customers, like an email marketing tool or a referral platform. The Model Context Protocol (MCP) is an open-source standard that enables AI assistants to connect to external tools and data.
MCP is a revolutionary development in the world of agentic marketing. Anthropic introduced MCP less than two years ago, on November 25, 2024. In Anthropic’s MCP announcement, the company called MCP “a new standard for connecting AI assistants to the systems where data lives, including content repositories, business tools, and development environments.”
In this guide, we’ll explore how marketers can use MCP-enabled tools to elevate their customer engagement strategies, including a rundown of what engagement platforms offer MCP support and how to choose the right one for your brand. We use the word “engagement” to refer to a range of marketing activities—in addition to messaging, it includes loyalty, referrals, and rewards that motivate customers to take action.
How MCP Works (and How It Differs from a Traditional API)
MCP works like a shared plug between AI tools and business software: the AI tool is the MCP client, and the platform it connects to is the MCP server.
An API (application programming interface) is a set of rules for how two programs connect and share information. With a traditional API, developers write custom code, called a wrapper, for each task. With MCP, the server lists its tools, and the AI client discovers and uses them on its own.
Not every MCP is set up the same way—the Intercom MCP server and the Klaviyo MCP server let AI tools work inside those platforms, while Zendesk offers a client, so its AI agents can use tools hosted elsewhere.
Across the B2B tech world, MCP availability has grown quickly. As of December 2025, Anthropic’s MCP adoption update reported: “There are now more than 10,000 active public MCP servers, covering everything from developer tools to Fortune 500 deployments.” In July 2026, MCP’s lead maintainers wrote in the MCP 2026-07-28 release announcement: “Across our Tier 1 SDKs, we’re seeing close to half-a-billion downloads a month, with both TypeScript and Python SDKs crossing the 1 billion total downloads threshold.”
So what does this mean for marketing and CX teams?
Why MCP Support Matters for Marketers
For marketers, the benefit is speed: a loyalty manager can ask an AI assistant for a performance report or reward change instead of filing a ticket. We’ve already seen the impact of AI on other roles, like developers and engineers, who already use AI tools to streamline their work. The Stack Overflow 2026 Developer Survey reported in October 2026 on developers’ AI use at work: “83% use at least one AI tool. Coding assistants (66%) and general chat tools (63%) dominate; agents/workflows reach 26%.”
Agents are also moving into customer workflows, and Gartner predicts that shift will grow. The Gartner agentic customer service forecast states: “By 2029, agentic AI will autonomously resolve 80% of common customer service issues without human intervention, leading to a 30% reduction in operational costs.”
Marketing and CX leaders are primed to be next on the list of teams who have their workflows revolutionized by AI, and MCP-enabled platforms are the tools to make that happen.
6 Customer Engagement Platforms with MCP Support
We selected six vendors who confirmed an MCP server or client in their official docs as of October 2026. Each entry covers what the MCP enables, the best fit industry, authentication methods (how a platform confirms who is connecting), and supported AI clients.
1. Extole: MCP for Loyalty, Referral, and Reward Programs
Extole is an enterprise offer management platform that powers referral, loyalty, reward, and incentive programs for enterprise brands. The Extole MCP server connects to any MCP-compatible AI environment, including Claude, Cursor, and ChatGPT.
Extole’s model is AI-assisted configuration and deterministic execution. Use AI to help build, track, and manage programs, while a rules-based runtime ensures your rewards are delivered with complete accuracy.
- What it enables: Run reports, analyze performance data, troubleshoot missing rewards, assist with campaign setup, get data-back suggestions, review participant history, reward history, campaign status, conversion data, and change logs.
- Authentication: OAuth with scoped tokens, so each action runs under the signed-in user’s existing permissions.
- Runtime: Extole handles sub-200ms event processing, real-time eligibility, live fraud detection, reward authorization, and fulfillment.
- Best fit: Brands across retail, telecom, financial services, and home services who want to run personalized incentive programs with AI speed and program integrity.
Extole uses OAuth to grant an app access to your Extole data without sharing a password, and a scoped token acts as a key that is limited to specific actions. See the full list of read and write MCP operations.
In one internal example, a complex SSO (single sign-on) integration that typically needs multiple review cycles took approximately 30 minutes with MCP-assisted configuration.
APIs still matter underneath; Every Extole MCP capability is backed by Extole’s published API platform.
2. Intercom: MCP for Support Conversations and Help Center Content
The Intercom MCP server is remote, meaning Intercom hosts it and AI tools connect over the internet. According to Intercom’s MCP developer docs, it offers 14 tools for support data.
- What it enables: Search and fetch conversations, contacts, and companies, create or update help center articles, and add internal notes.
- Authentication: OAuth, which Intercom recommends, or a Bearer token, an access key sent with each request.
- Regions: US and EU endpoints, the web addresses tools connect to, are available; AU is not yet supported.
- AI clients: Setup guides cover Claude Desktop, Claude Code, OpenAI, Cursor, Windsurf, and VS Code.
- Best fit: Support teams that want AI tools to search conversations and keep help content current.
3. Braze: MCP for Cross-Workspace Lifecycle Marketing
Braze announced its MCP Server as generally available in its MCP Server announcement article, published July 23, 2026. It connects Braze to Claude, ChatGPT, and Cursor.
- What it enables: Natural-language data queries, so teams can ask questions about their data in plain English.
- Cross-workspace work: Migrate assets between Braze workspaces through an AI assistant.
- Mobile access: Use the server from a phone or tablet through remote MCP.
- Authentication: Built on OAuth standards.
- Best fit: Lifecycle marketing teams that manage several Braze workspaces.
4. Klaviyo: MCP for Email and SMS Campaign Workflows
Klaviyo MCP gives email and SMS marketers an AI path into their campaigns. The Klaviyo MCP server lets AI clients help with campaign building and reporting, per Klaviyo’s MCP server guide.
- What it enables: Tasks such as campaign creation and flow performance reports, where a flow is an automated message series.
- Setup options: A remote-hosted server, which Klaviyo recommends, or a local server on your own machine.
- AI clients: Clients such as Claude or Cursor.
- Best fit: Ecommerce teams running email and SMS programs in Klaviyo.
5. Salesforce Marketing Cloud Engagement: Hosted MCP Server
Salesforce Marketing Cloud Engagement offers a hosted MCP server, so Salesforce runs it for you. Setup uses an installed package plus an API Integration component, as the Marketing Cloud MCP setup guide explains.
- What it enables: Help managing customer data, orchestrating campaigns, and coordinating data transfers.
- Regions: US and EU servers.
- AI clients: AI assistants such as Claude Code or Gemini CLI.
- Best fit: Organizations that already run marketing in Salesforce.
6. Zendesk: MCP Client for AI Agents and Action Flows
Zendesk works from the client side of MCP. Its MCP client became generally available on August 10, 2026, according to Zendesk’s general availability announcement. It lets Zendesk’s AI use tools from other systems.
- What it enables: Admins connect public and private MCP servers, and AI agents and auto assist use those tools in action flows (step-by-step task sequences).
- Plans: Suite Growth plans and above, and Support Team plans and above.
- Best fit: Support teams that want AI agents to take action in other systems.
Side-by-Side Comparison
Use this table to narrow your list of tools before you test anything.
| Platform | MCP role | Primary use | Authentication | AI clients named | Regions or plan notes |
|---|---|---|---|---|---|
| Extole | Server | Loyalty, referral, and reward programs | OAuth with scoped tokens | Claude, Cursor, ChatGPT | Actions run under user permissions and are logged |
| Intercom | Remote server | Support conversations and help content | OAuth or Bearer token | Claude, OpenAI, Cursor, Windsurf, VS Code | US and EU; AU not yet supported |
| Braze | Server | Cross-workspace lifecycle marketing | OAuth standards | Claude, ChatGPT, Cursor | Mobile access via remote MCP |
| Klaviyo | Remote or local server | Email and SMS campaigns | See Klaviyo docs | Claude, Cursor | Remote-hosted recommended |
| Salesforce Marketing Cloud Engagement | Hosted server | Customer data, campaigns, data transfers | OAuth via installed package / API Integration | Claude Code, Gemini CLI | US and EU servers |
| Zendesk | Client | AI agents and action flows | Admin-managed connections | Not applicable (client) | Suite Growth+ and Support Team+ |
MCP Security and Governance: What to Check
AI security comes down to one question: what exactly can the AI do, and who answers for each change? The MCP authorization specification explains the model: “A protected MCP server acts as an OAuth 2.1 resource server, capable of accepting and responding to protected resource requests using access tokens.”
The spec makes authorization optional, so confirm how each vendor handles it. Use this checklist when you review any MCP server.
- Scoped tokens: Access should cover only the actions a task needs.
- User-level permissions: AI actions should run under the person’s existing permissions.
- Change attribution: Every change should record the user, the tool, and the time; Extole’s change log captures all three.
- Safe testing: Look for a way to test programs in a sandbox before changes reach customers.
- Runtime separation: AI should edit setup while a deterministic engine runs live rewards and fraud checks.
Risk control is critical as it can also be the deciding factor in whether AI projects survive. The Gartner agentic AI cancellation forecast states: “Over 40% of agentic AI projects will be canceled by the end of 2027, due to escalating costs, unclear business value or inadequate risk controls….” This is not an inevitability; careful assessment of the AI tools you use can help you identify possible security gaps before they emerge.
Another factor to consider when it comes to reward programs is fraud. Our guide to enterprise fraud prevention requirements covers the questions to ask and features to look for, such as manual review gates for larger rewards, or audit trails for highly regulated industries.
How to Choose the Right Platform
Use these seven steps to compare customer engagement platforms with MCP support. Score each platform against your top use case first.
- Use-case fit: Match the platform to the job, such as support, lifecycle messaging, email and SMS, or offers and rewards.
- Read vs. write depth: Check which actions AI can take beyond reading data, especially changes that touch money or customers. Some capabilities may still be in development—make sure to clarify what is live and what is still in beta.
- AI client compatibility: Confirm the MCP can connect to the tools your team uses, such as Claude, ChatGPT, Cursor, or Gemini CLI.
- Security and auditability: Ask about OAuth, scoped tokens, user permissions, and change logs.
- Configuration vs. runtime separation: Ask where AI exists in the process, and what runs as rules-based.
- Plan and region requirements: Verify plan tiers and data regions, such as Zendesk’s plan limits or Intercom’s endpoints.
- Team capacity: Choose a hosted server for a fast start, or a local one if developers can manage it.
For a wider view of evaluation criteria, read our guide to agent-ready marketing platforms.
FAQ
Here are quick answers to common questions about MCP options.
What Is a Customer Engagement Platform?
A customer engagement platform is software, such as a CRM, messaging, support, or offer management tool, that helps brands reach, support, and reward customers.
What Is MCP?
The Model Context Protocol is an open-source standard, introduced by Anthropic in November 2024, that connects AI assistants to business tools and data.
Can I Use ChatGPT or Claude with These Platforms?
Braze and Extole name Claude, ChatGPT, and Cursor, and Intercom provides setup guides for Claude, OpenAI, Cursor, Windsurf, and VS Code. Salesforce names Claude Code and Gemini CLI, so check each vendor’s list before you connect.
How Does MCP Differ from an API?
A traditional API needs custom code for each task, while MCP lets an AI client discover a server’s tools and use them directly.
What Authentication Do MCP Servers Use?
Intercom supports OAuth or Bearer tokens, Braze is built on OAuth standards, and Extole uses OAuth with scoped tokens. The MCP specification defines OAuth 2.1-based authorization but makes it optional.
Which Zendesk Plans Support the MCP Client?
The Zendesk MCP client is available on Suite Growth plans and above and Support Team plans and above.
Ready to invest in a customer engagement platform with MCP support? Extole combines AI-enabled referral, reward, and loyalty tools with secure reward delivery. Book a demo to see what you can build with Extole MCP.
