Grok BotUndisclosed

Talent team runs 7-day recruiting outbound with multi-bots; Grok Bot suggests tool to unblock sends

ericosiuX (Twitter) · 2026-11-06

Read original

Our talent bots run recruiting outbound 7 days a week from 5 sourcing channels (and any campaign that bounces over 2% pauses itself) I broke down how it works ↓ I built it myself because if you don't define how talent works at your company as a leader, your talent team will define it for you 3 bots run it from Slack: Team Talent, Team Juicebox and Team LinkedIn Recruiter. They pull candidates from 5 places: > @teamtailor, our applicant tracking system > posts on X, which bring in a ton of applicants > LinkedIn Recruiter InMails > Juicebox, an agentic recruiter > @InstantlyAI, which is cold email infrastructure Then every candidate goes through the same steps: 1. check our do-not-contact list, like friendly companies and people we've turned down before 2. check title fit 3. for mid-senior roles, look for 2 promotions, ideally at 2 companies, and about 10 years at an average tenure of 3+ 4. send a first message, capped per role 5. verify every email, drop catch-alls and auto-pause any campaign over a 2% bounce rate 6. send the 7 questions to anyone interested 7. Jev scores the answers A funnel report lands in Slack every day: pre-qualified, emailed, replied, interested, 7 questions sent, answered within 24 hours, shortlisted Step 5 is the part a marketer brings. Most talent people have never had to think about deliverability, and a burned inbox takes the whole campaign with it The first report showed every role running under its daily cap. Juicebox was the bottleneck, because sending more meant paying for more inboxes So I bought inboxes on Instantly and warmed them up, and Grok @bot suggested Treg to find personal emails. Treg is 60 to 70% accurate, which is enough to stop the sends from being the bottleneck

Business model

Founder ericosiu runs talent bots doing recruiting outbound 7 days a week from 5 channels, auto-pausing any campaign over 2% bounce. When sends became the bottleneck, Grok Bot suggested Treg to find personal emails (60-70% accurate), which was enough to remove the limit. Readers can reuse using an Agent to diagnose ops bottlenecks and recommend tools.

Workflow

  1. Pull candidates from 5 sources and pre-screen them
  2. Send first messages (capped), verify emails, auto-pause campaigns over 2% bounce
  3. When sends bottleneck, use Grok Bot to diagnose and suggest a tool

Adaptation plan

Funnel details and the Grok suggestion are as reported by the original post author. Full steps and Prompt are AgentFlow reconstructions and were not live-tested here. No income claimed. Grok Bot and related tool subscription amounts undisclosed; full multi-bot workflow details not shown. Cold outreach with personal-email lookup carries compliance and deliverability risk; verify on a small batch first.

Prompt

AgentFlow draft · Untested.

You are my recruiting ops advisor. Diagnose the current outbound bottleneck from the materials I provide:
- Campaign name and daily send cap
- Current bounce rate and pause rule
- Candidate sources and verified-email ratio

Task: Identify the main bottleneck, suggest 1-2 actionable improvements (tool or process), and note risks.
Output: Concise Markdown with diagnosis, suggestion list, confirmation points.
List gaps if data is missing; do not invent accuracy rates or numbers. Wait for my confirmation before any suggested send or purchase.
Untested AgentFlow reconstruction.

Running costs and pitfalls

Funnel details and the Grok suggestion are as reported by the original post author. Full steps and Prompt are AgentFlow reconstructions and were not live-tested here. No income claimed. Grok Bot and related tool subscription amounts undisclosed; full multi-bot workflow details not shown. Cold outreach with personal-email lookup carries compliance and deliverability risk; verify on a small batch first. No independent cost benchmark is available. Check subscriptions, permissions and external actions before running a workflow.

Author-reported results are not independently audited.

Original text and media remain attributed to their authors.