How one engineer saved ClickUp’s support team 917+ hours a month

Corey Smith is a Senior Technical Support Engineer at ClickUp. His team handles about 5,000 support tickets a month, and each one used to start with 15 minutes of research before a rep could type a reply. Corey built a system that cut that to 4 minutes. Across 5,000 tickets, that's roughly 917+ hours back every month.

Before every customer ticket, a support rep at ClickUp had to pull context from Zendesk, cross-reference internal documentation, and match it to the right help article or runbook. About 15 minutes per ticket. Roughly 5,000 tickets a month.

Corey had tried traditional Zapier automations to speed this up: triggers and actions wiring Zendesk to other tools. That covered the straightforward stuff. But the workflow he actually needed required pulling unstructured ticket data, running it through AI for interpretation, then routing the output into something a rep could act on. You can't chain that kind of workflow together with individual Zaps.

Corey's setup runs on Zapier MCP. When a ticket lands in Zendesk, his system pulls the full context and maps it against ClickUp's internal knowledge base using AI. The rep gets a structured summary before they type a word: relevant docs, recommended steps, and any related past tickets.

The same setup handles the back end too. After a resolution, Corey's system pulls feedback data and review summaries so team leads can see how tickets are landing without running manual audits.

Research time dropped from 15 minutes to about 4. Other teams at ClickUp noticed and started asking for the same thing.

Corey describes MCP as API access you don't have to build yourself. If you've read API documentation and thought "I could do something with this if I had the time to wire it up," that's the gap. You get structured access to Zendesk, Salesforce, Google Workspace, and hundreds of other tools without writing custom integrations. MCP connects you to the data, and the AI figures out what to do with it.

The shift from traditional automation is conceptual more than technical. Zaps work well for predictable, repeatable flows. When the input is unstructured, like a freeform customer ticket or a messy email thread, you need interpretation before routing. That's what Zapier MCP and an AI layer handle together.

Corey's MCP adoption advice: concrete examples beat documentation. Show people what you're connecting to and what data comes back. Most people who haven't tried MCP picture "build your own integration," which sounds hard. When he reframed it as "structured API access," the other teams at ClickUp got interested.

Reps on Corey's team start every ticket with context already in hand. Docs surfaced, history pulled, response path laid out. A 15-minute research cycle became a 4-minute processing step.

Corey's team chose Zapier MCP to streamline ClickUp's Support Operations when the problem needed interpretation on top of automation.

Corey's advice for getting started was simple: look at concrete examples first. If you want to see what MCP can do, browse the MCP template library and pick one that matches what you're trying to get done.

Corey Smith is a Senior Technical Support Engineer at ClickUp,