Zapier Shares How Customer Advocacy Team Built AI-Native System With MCP and SDK.

Zapier has shared a detailed look at how Rob Ayre, a customer advocacy team member, rebuilt the company's entire customer advocacy function as an AI-native operating system. The build, announced in a LinkedIn post by Lindsay Rothlisberger, showcases a practical application of Zapier's SDK and Model Context Protocol (MCP) integration, tools the company recently opened up to builders.

Ayre joined Zapier just a few months ago and used Cursor, the Zapier SDK with MCP, Databricks MCP, Zapier automations, and Zapier Tables to replace what Rothlisberger described as the typical manual process: one person managing a spreadsheet and a chain of direct messages.

The system Ayre built includes five core components. First, a Slack-based intake process that logs potential customer advocates into the pipeline automatically when anyone on the team adds an emoji reaction to a relevant message. This eliminates the need for forms or direct messages to the advocacy team.

Second, an advocacy CRM now tracks 230 accounts in an enriched pipeline. Third, a story discovery feature uses Cursor to pull real account and CRM data, surfacing which customers have stories worth telling and suggesting narrative angles. Fourth, a content library houses more than 700 stories, quotes, statistics, and slides, all formatted and approved for use.

The fifth component is a pre-call proof brief. When a sales representative posts a company name in Slack, the system generates a full page in Notion or Coda within minutes, complete with peer companies, published case studies, talk tracks, and relevant statistics. According to Rothlisberger, this process previously required more than 30 minutes of manual research.

Ayre manages the entire system through a single Cursor conversation, a workflow that demonstrates the practical application of the Zapier MCP integration the company launched earlier this year.

The system also includes a Weekly Pipeline Pulse that posts automatically to the team channel every Wednesday at 8am. The update includes all stories shipped, interviews booked, new inbound leads, and new contacts spotted. Ayre reviews a draft, replies with a simple "go," and the update publishes. The entire team stays informed about the state of the customer story pipeline without requiring standup meetings.

Rothlisberger positioned the build as an example of what an AI-native hire looks like in practice: someone who uses AI not just to work faster, but to build a fundamentally different version of the job. She noted that customer advocacy has historically been one of the highest-leverage, most under-resourced parts of a go-to-market motion.

The technical foundation of Ayre's system draws heavily on tools Zapier has been developing and opening up to users. The Zapier SDK allows builders to replace multi-step automations with custom code, while the MCP integration lets AI tools like Cursor perform thousands of actions across connected apps without complex API integrations.

Databricks MCP, an open-source Model Context Protocol server, enables AI applications to interact with data platforms, query databases, and retrieve analytics. Combined with Zapier Tables for data storage and Zapier automations for workflow orchestration, the stack creates a system that runs largely on autopilot once configured.

The post included an open question to the LinkedIn audience about how other companies currently operate their customer advocacy or reference programs, with Rothlisberger suggesting that Ayre would be willing to share more details with teams still running manual processes.