Claude
ChatGPT
Gemini
CoPilot

Run your hiring program by asking.

The conversational interface to your entire Talkpush platform. Ask any question or take any action, in plain language.

Request access
mcp.talkpush.com
Loop 01 of 03 · Diagnose
mcp.talkpush.com
Loop 02 of 03 · Act
mcp.talkpush.com
Loop 03 of 03 · Set up

A conversational interface for your entire hiring funnel.

"By the time the dashboard catches the issue, the campaign is two weeks behind."
Diagnose program-level drift the same week it happens, not the next quarter.
See why your offer rate, time-to-hire, or channel mix is moving, without waiting for a slice.
Walk into the CHRO review with a live answer, not a screenshot of last Friday's dashboard.
"I have 50 campaigns running. I can't tell which autoflows are misfiring until a recruiter complains."
Audit autoflow rules across every campaign at once. Surface the ones over-filtering before a recruiter does.
Pause, soften, or change a screening rule from the same thread you used to find it.
Iterate campaigns the same day. Never wait on a config audit again.
"I rebuild the same dashboard every quarter for the same questions. Then someone asks a follow-up and the rebuild starts again."
Answer ad-hoc questions in minutes, across stage, channel, recruiter, time, and cohort.
Drop the dashboard rebuild cycle. Save the patterns you keep finding to the team library instead.
Walk into the leadership review with answers, not slides.

Every question. Every action. One prompt away.

You
Show me candidates who went quiet after autoflow message #3.
AI assistant · Talkpush MCP
247 candidates stalled after message #3 in the last 14 days. 62% are concentrated in Manila BPO (94) and Bogotá retail (61). Mean time to last response: 9.2 days. Want me to queue a re-engagement send, or surface the recruiters who own them?
You
Which screening questions are filtering the most candidates this month?
AI assistant · Talkpush MCP
Top three this month: "Night-shift availability" (Manila, 62%), "Minimum tenure 6 months" (Bogotá, 58%), "Fluent English written + spoken" (Mumbai, 54%). The Bogotá rule reduces applicant volume 41% week over week but hasn't changed in 3 months. Want to see candidates these rules filtered who otherwise matched the role?
You
How is our newest voice agent performing vs. the previous version?
AI assistant · Talkpush MCP
Over the last 14 days: v4 ran 387 calls, 84% completion, average score 6.8. v3 ran 412 calls, 71% completion, average score 6.2. v4 is ahead on completion and score, slightly behind on volume (a 3-day pause explains it). Promote v4 to default?
You
Show me candidates who went quiet after autoflow message #3.
AI assistant · Talkpush MCP
247 candidates stalled after message #3 in the last 14 days. 62% are concentrated in Manila BPO (94) and Bogotá retail (61). Mean time to last response: 9.2 days. Want me to queue a re-engagement send, or surface the recruiters who own them?
You
Run a resume-match analysis for our top 50 applicants against the night-shift CS role.
AI assistant · Talkpush MCP
Match analysis complete. 14 of 50 score 80%+ against the JD (strong CS history, night-shift availability confirmed). 22 score 60–79% (partial fit). 14 score below 60%. The top 5 are ready to advance to screening. Want the full breakdown, or just the top picks?
You
Rank candidates within 15km of the Cebu hub for the on-site agent role.
AI assistant · Talkpush MCP
47 candidates within 15km of the Cebu hub. Top 10 are concentrated in Lapu-Lapu and Mandaue. Average commute estimate: 18 minutes. 6 of the top 10 also have prior CS experience. Want me to fast-track those 6 to screening?
You
Pause the autoflow rule filtering night-shift applicants in Manila for 72 hours.
AI assistant · Talkpush MCP
Done. Rule "night-shift availability" paused in the Manila BPO campaign until Friday 9am. The 184 applicants currently held by this rule have been moved back to the screening queue. Want me to message them?
You
Move stalled candidates over 14 days to the re-engagement queue.
AI assistant · Talkpush MCP
312 candidates match (stalled 14+ days at the application stage). Moving them to the Re-engagement queue and tagging "stalled-14d". Auto-send the standard re-engagement WhatsApp in the next batch? It runs at 10am local in each candidate's region.
You
Schedule screening interviews with the top 20 applicants from yesterday.
AI assistant · Talkpush MCP
Top 20 by TalkScore queued. Each candidate gets a WhatsApp message with three time options in the next 48 hours, mapped to your team's working hours per region. Confirmations sync to your calendar. Want me to flag the top 3 to specific recruiters?
You
Set up a new campaign for night-shift CS agents in Cebu, 200 hires, start Monday.
AI assistant · Talkpush MCP
Campaign drafted. Pulled the role template from your existing Cebu CS campaign. Channels: Facebook, JobStreet, internal referrals. Autoflow includes 6 screening questions and a TalkScore voice assessment. Launch it, or hand to RecOps to review first?
You
Create a voice agent that screens for English fluency and night-shift availability.
AI assistant · Talkpush MCP
Draft created. Prompt set to screen for B2-level conversational English and night-shift availability. Data extraction: fluency rating, availability confirmation, preferred shift. Publish now, or test on the last 20 applicants first?
You
Clone the Manila autoflow set into the Bogotá campaign and adjust the language to Spanish.
AI assistant · Talkpush MCP
Autoflow set cloned into Bogotá. 11 rules carried over. Message templates switched from Tagalog to Spanish. Time zone shifted to America/Bogota. Three rules flagged for review since they reference Manila-specific channels. Want me to surface those?

Read, act, configure, build.

Anything a recruiter, RecOps lead, analyst, or admin does inside Talkpush is callable through MCP.

01
Campaign Discovery
6 tools
02
Campaign Management
5 tools
03
Candidate Search & Details
5 tools
04
Candidate Actions
8 tools
05
Candidate Attributes
2 tools
06
Labels
2 tools
07
Voice Agents
7 tools
08
Calls
1 tool
09
Messaging
6 tools
10
Autoflow Discovery
3 tools
11
Autoflow Management
9 tools
12
Question Discovery
5 tools
13
Question Management
12 tools
14
Company Discovery
2 tools
15
Company Folders
2 tools
16
Store & Location
3 tools

An open layer between your platform and any AI.

Use whichever AI your team prefers. Switch later without rewriting anything.

Five minutes. No engineering.

Configuration, not integration. Generate credentials, paste them into your AI client, start asking.

Step 01
Generate credentials.

In the Talkpush admin panel, on the Integration Setup page, generate a client_id and client_secret.

Step 02
Paste into your AI client.

Claude, ChatGPT, Gemini Enterprise, or your custom agent runtime.

Step 03
Ask anything.

Start with the question your last dashboard couldn't answer. Or the action your last ticket is still waiting on.

Enterprise-grade by default.

The AI inherits each user's role and can't act beyond it. Every prompt and response is logged for your IT team to audit.

OAuth 2.0 + PKCE

Authorization using user-specific, revocable credentials issued per user.

Three role tiers

talkpush_owner, company_admin, company_user.
The AI inherits the user's role. It never bypasses it.

99.9% uptime target

Serverless, auto-scaling, globally distributed. Rate-limited at 60 requests per minute per company.

Signed session tokens

Session tokens are signed with HMAC-SHA256. Tampering is detected on every request.

Tenant isolation

Enforced by the credentials scope. Every query filtered by company at the protocol layer.

100% audit coverage

Every prompt and every response logged with full input and output payloads.

PII fetched on demand

Candidate PII is never stored in the MCP layer. AI providers operate under your existing data agreements.

Redirect URI allowlisting

OAuth redirect URIs must be on the approved allowlist. No unbounded redirect surface.

FAQs

Frequently Asked Questions

How is this different from a BI dashboard or a scheduled report?
Letting an AI act on our recruiting data sounds risky. How does this work?
Is the AI actually good enough to interview candidates?
Is this free for Talkpush customers?
How quickly can we go live?
What can I actually ask it to do?
How long does it take to set up?
Does this work with our existing ATS, HRIS, and BI tools?
What if we want to switch from Claude to Gemini or ChatGPT later?
What changes for my team if we start using this?

Try it. It's already running.

Built for scale and security
at every step

Our platform employs end-to-end encryption and multi-factor authentication, providing a secure environment for your recruitment data.

Loved by people
around the globe

We proudly serve our customers to the highest level.

Follow our events and product updates to stay informed about the latest advancements in recruitment technology and industry trends.

All fields are required

Thank you for registering!

You’re now part of our community. We’ll keep you updated on upcoming events, new product releases, and the latest trends in recruitment technology. Stay tuned for exclusive insights and opportunities delivered straight to your inbox.