How a Recruiting Agency Uses Claude and Zapier MCP to Find Candidates Who Aren't Looking

Most good hires aren't sitting on job boards. They're already employed, not checking LinkedIn, and would never reply to a mass InMail. The only way to reach them is to notice they exist, usually buried in a CRM record nobody has opened in eight months, and say something worth replying to. That's the exact problem a recruiting agency called Blue Castle Agency is reportedly solving by wiring Claude into Zapier MCP, letting an AI assistant read and act inside the tools the agency already uses instead of building a custom integration from scratch.

In recruiting, there are two pools of people: active candidates who are applying, and passive candidates who are employed, not searching, but open to the right conversation at the right time. Agencies have always known the second pool is where the strongest hires come from. The problem has never been strategy, it's been labour. Someone has to keep old contacts warm, notice when a person's situation might have changed, and write a message that doesn't read like a template. That's exactly the kind of repetitive, judgement-light research work an AI model paired with action access can take off a recruiter's plate.

Curtis Key has spent roughly 20 years building automation, starting with basic code and scripts before moving into Salesforce automation at a real estate brokerage that grew from 8 agents to 105. When hiring became that brokerage's hardest operational problem, Key founded Blue Castle Agency, a recruitment firm focused exclusively on real estate and property management roles. That niche focus matters: a generalist recruiter can't build deep pattern recognition across thousands of candidate profiles, but a firm working one industry for years can teach an AI system exactly what a 'good fit' looks like for that world.

MCP stands for Model Context Protocol, an open standard that lets an AI model like Claude talk to outside applications in a structured way. Zapier MCP is Zapier's implementation of that standard: it sits between an MCP-compatible AI client and your Zapier account, and according to Zapier it currently exposes access to more than 9,000 apps and more than 40,000 actions. Zapier handles the messy parts, app connections, stored credentials, and rate limits, so the AI model never has to know how Salesforce's API or Gmail's API actually works. It just asks for an action, and Zapier executes it.

According to Zapier's own documentation, Claude connects to Zapier MCP through a connector, and the person setting it up chooses exactly which Zapier actions Claude is allowed to touch. That's a meaningful detail. You're not handing Claude your entire CRM, inbox, and calendar in one shot. You're saying: Claude can read contact records in this one list, it can draft (not send) emails through this one account, and it can create calendar events in this one calendar. Zapier's own examples of what this kind of connection supports include reading Salesforce records, posting messages to Slack, and creating Google Calendar events, all without anyone writing integration code.

Here's where it's important to be straight with you. The case study describing Blue Castle's workflow carries a publish date of October 22, 2026, one day after this research was gathered, which means it cannot be treated as fully published or independently checked yet. There are no verified numbers available for candidate volume, placements made, hours saved, or cost, and we're not going to invent any. What we can say with confidence is the pattern: Key's background is in building automation inside existing systems rather than replacing them, first in raw code, then inside Salesforce, and now reportedly inside a Claude plus Zapier MCP pipeline. That continuity, automate inside the tools you already trust rather than ripping them out, is the part worth copying even before the full case study is public.

Based on how Zapier MCP and Claude are documented to work together, here's an illustrative version of how a recruiting agency could build this, not a confirmed Blue Castle blueprint, but a realistic one using the same building blocks.

Zapier has also been rolling out something called Next Gen Zaps, where you describe a workflow in plain language to an MCP-compatible AI client like Claude, ChatGPT, or Cursor, and it builds the automation for you. As of October 19–20, 2026, Zapier describes this as limited availability and still in active development, so it's early, but it points at exactly this kind of use case: a non-technical founder describing a sourcing workflow in a sentence instead of configuring twelve separate Zap steps by hand.

A sourcing pipeline that only waits for people to apply is only sourcing from people who already quit their last job.

The mechanics are documented and genuinely accessible, but most small teams trip on the same handful of things.

There's no India-specific data in the research behind this story, so take this as a reasonable inference rather than a reported result. But the underlying capability doesn't care which country it runs in. An Indian staffing agency with a Zoho CRM full of candidates who went quiet two years ago, or a B2B sales team with a HubSpot pipeline full of leads nobody re-engaged, has the exact same problem Blue Castle is reportedly solving: good records sitting unused because nobody has the hours to work them manually. If your outreach actually happens over WhatsApp rather than email, which is how most Indian B2C and B2B follow-up genuinely gets read, ODIV Engage's CRM and shared inbox can sit on the receiving end of that same Zapier-and-Claude pipeline so replies land somewhere your team actually watches.

This is exactly what ODIV's ai-workflow-automation service is for: taking a business process that currently depends on one person's memory and scattered spreadsheets, and wiring it into Claude, Zapier, or similar tools with proper scoping, human-review checkpoints, and ongoing ownership, so it keeps working after launch instead of breaking in week three. For a recruiting firm, sales team, or any founder sitting on a CRM full of dormant contacts, ODIV's engineers would map your actual sourcing or follow-up workflow, build the scoped Claude-to-Zapier connection, set the right approval gates, and hand you a system you can actually trust. Our team works hands-on in AI build tools like Lovable and Claude Code alongside conventional engineering, which is precisely why this kind of build lands in a fraction of the time and cost of a traditional hand-coded custom development project, you're not paying someone to write integration code line by line, you're paying for judgement on top of tools that already move fast. If you've got a pile of contacts nobody's working and a team too small to work them manually, start a chat with ODIV on WhatsApp and we'll tell you honestly what a workflow like this would take to build for you.