Skip to main content

Getting Started with Talkpush MCP

Everything you need to know about the Talkpush MCP — what it is, what it can do, and how to connect it to your AI platform.

Written by Crismin Joy Lagamayo

The Talkpush MCP (Model Context Protocol) connects your AI assistant directly to Talkpush — giving it live access to your campaigns, candidates, autoflows, and more. Instead of switching between tabs or running reports manually, you ask your AI agent a question and it pulls the answer straight from your data.

This article covers what the Talkpush MCP is, why it matters, and how to get your credentials before connecting to your preferred AI platform.

What is MCP?

The Model Context Protocol (MCP) is an open standard, originally introduced by Anthropic, that lets any LLM client talk to any external tool or data source through a uniform interface. In practice, this means AI assistants like Claude, ChatGPT, or Gemini can take actions inside Talkpush on your behalf — not just answer questions, but actually search candidates, update campaigns, trigger autoflows, and more.

MCP amplifies what you can do with Talkpush

Talkpush MCP goes beyond the Talkpush UI. While the UI is built for humans to work screen by screen, MCP allows AI assistants to work across campaigns, candidates, autoflows, configurations, and analytics all at once.

This means you can process work at a much larger scale, connect insights across your hiring program instantly, and generate real-time analysis tailored to the exact question you're asking — without being limited by tabs, dashboards, or predefined workflows.

Every tool at a glance

The Talkpush MCP gives your AI assistant access to 61 tools across 12 categories: Campaign Discovery, Campaign Management, Candidate Search & Details, Candidate Actions, Autoflow Discovery, Autoflow Management, Candidate Attributes, Labels, Voice Agents, Messaging, Store & Location, and Company Folders. The AI uses these autonomously based on your natural language requests.

Below is a breakdown of all available tools organized by category, giving you a clear view of what the MCP can do at a glance.

Click any tool at mcp.talkpush.com to see the endpoint, parameters, and example prompts.

Actions only Talkpush MCP makes possible

Each one is possible because of what Talkpush MCP unlocks: processing many records in a single interaction, pulling data across the whole hiring program at once, reshaping analytics on demand, or running AI-driven analysis on top of it all.

Recruiter

Action

What it does

Sample prompt

Evaluate and rank multiple candidates simultaneously against custom criteria

Score and rank a set of candidates against role-specific criteria you define on the fly. Replaces hours of manual review with a single conversation; the assistant evaluates every candidate against the same rubric and returns a ranked list with reasoning.

Take the top 20 screened candidates in the [campaign name] campaign and rank them by [criterion 1] and [criterion 2]. Explain the ranking briefly.

Run AI-driven position-fit analysis on candidate resumes and LinkedIn profiles

Pull a candidate's resume or LinkedIn profile and get a structured, AI-driven fit analysis against the job posting. The assistant surfaces matches, gaps, and red flags in seconds, instead of opening and reading every document by hand.

Pull the resume and LinkedIn profile of [candidate name] and tell me how they stack up against the requirements for the [campaign name] campaign.

Detect silent candidates and re-engage them

Identify candidates that have gone quiet and act on them in the same conversation. The assistant flags candidates by stage and last activity, then sends a follow-up message so nobody drops out unnoticed.

List candidates in the Shortlisted folder of the [campaign name] campaign who haven't been updated in more than 7 days, and send them our follow-up message.

Rank candidates by geographic proximity to a location

Sort candidates by how close they live to a specific location (coordinates). The assistant resolves the location, fetches each candidate's address, computes distances, and returns a ranked shortlist.

From the screened candidates in the [campaign name] campaign, give me the 5 closest to [coordinates], sorted by distance.

Configuration Owner

Action

What it does

Sample prompt

Configure multiple autoflows from a single brief

Create or update several autoflows in one go, with shared logic or per-campaign overrides. No clicking through the UI rule by rule; describe the behavior once and the assistant applies it across every campaign you point to.

For every active campaign in [region or segment], set up an autoflow that auto-moves candidates to the Shortlisted folder when they [trigger condition].

Create or update multiple campaigns at once with shared configuration

Launch or modify many campaigns in a single conversation using a shared baseline (job description, screening questions, autoflow set) and per-campaign overrides (location, headcount). Useful when rolling out the same role across regions.

Create 5 [role] campaigns, one per store in our [region], all using the same job description and screening questions but with the location set per store.

Clone an entire autoflow set with adjustments

Duplicate a working autoflow set onto another company or tenant and apply targeted edits in the same step (swap folder IDs, update messages). Roll out proven automation to new markets without rebuilding it from scratch.

Clone the [autoflow set name] autoflow set from [source company] onto [target company], swap the folder IDs to match the target, and update each message template to its equivalent in the target company.

Audit every campaign for misconfigurations

Scan campaign configurations for inconsistencies, missing fields, missing autoflows, or anything that deviates from your standards. The assistant flags anomalies and proposes the fix.

Audit every active campaign and tell me which ones are missing the On-site Interview folder, have no autoflows configured, or have no salary information.

Diagnose funnel health across campaigns in real time *

Get a live, cross-campaign view of how the funnel is performing right now. Stage-by-stage conversion, where candidates are stuck, where they're dropping out. No waiting for tomorrow's dashboard refresh.

Give me a current view of the funnel for every active campaign in [region or segment]. Where are we losing the most candidates and at which stage?

Diagnose root causes from multiple angles when metrics shift unexpectedly *

When a metric moves in a way you didn't expect, trace where in the program the change is happening and why. The assistant investigates from multiple angles (campaigns, autoflows, candidate activity, channel mix) against live data, in real time, instead of waiting for a postmortem.

Candidate volume dropped 10% week over week on the [region or segment] campaigns. Help me figure out where the drop is happening and what changed.

* These actions are also applicable to Hiring Program Owners.

Connect your AI platform to Talkpush

The Talkpush MCP works with the AI platforms your team already uses — including Claude, ChatGPT, and Gemini. But before you can connect any of them, you need an account on the Talkpush MCP portal. This is where your credentials live, and without them, none of the platform setups will work.

To get access, you need to be invited to the Talkpush MCP. You'll receive an email with a link to log in at https://mcp.talkpush.com/. If you haven't received an invitation yet, contact your account manager.

Once logged in, go to the Integration Setup tab to find your MCP endpoint and your credentials: the Client ID and Client Secret. Click on them to copy to your clipboard. You'll need these in the next step when connecting to your platform of choice.

Follow the setup guide for your platform:


Did this answer your question?