Agents In Use
What’s new todayLast update
3 use cases added today

Grok Bot drafts Shopify outreach, but misjudges lead size

@malmaladei reports that Grok Bot researched, scored and drafted emails for 20 prospects, but 17 were too large. He added approval before sending, explicit disqualifiers and an unclear flag for uncertain signals.

Source published x.comHow we sourceAgent: Grok Bot
OUR ADAPTATION

Suggested prompt

Suggested prompt written by Agents In Use, not the source author. Replace bracketed inputs and review tool permissions first.

Research German Shopify stores with roughly €20–100k monthly revenue and complex products. State the evidence for fit, apply my disqualifiers and mark uncertain signals unclear. Draft outreach, but wait for my approval before sending.

Evidence at a glance

Sourced account
Source poster / publisher
@malmaladei
Original account
Read the source ↗
Published
Source link checked
Reported result
@malmaladei reports 20 drafted emails, with 17 of 20 leads too large after re-checking. He introduced approval before sending and stricter qualification rules; no results from the revised workflow are reported.
Human involvement
Hoffmann set the target, spotted an oversized prospect and requested the re-check. He now requires approval in chat before any batch is sent.

A linked account supports attribution. It does not mean we independently reproduced its result.

Report an issue with this caseOpen correction email

Opens your email app with the case link included. Nothing is sent automatically.

How the agent did it

  1. Hoffmann briefed Grok Bot to find 10 German Shopify leads at roughly €20–100k monthly revenue with complex products.
  2. Grok Bot researched and scored prospects and drafted 20 personalised emails.
  3. Hoffmann noticed an oversized brand and asked the bot to re-check; 17 of 20 leads were too large.

Philipp M. W. Hoffmann, posting as @malmaladei, asked Grok Bot to find 10 leads for his startup. His target was German Shopify stores with roughly €20–100k in monthly revenue and complex products likely to generate support tickets. He wanted early customers he could onboard personally.

He reports that the bot researched prospects, scored them and drafted 20 personalised emails in minutes. The problem emerged when he noticed a multi-million-euro brand labelled as mid-sized. After he requested another check, 17 of the 20 prospects turned out to be too large. His explanation is that the bot used brand visibility as a revenue signal, favouring businesses that were easier to find rather than ones that fitted the brief.

Hoffmann changed the workflow to keep research and drafting autonomous while requiring approval before sending. He also added explicit disqualifiers and instructed the bot to mark uncertain signals as unclear rather than guess.

This is an account of a failed qualification pass and the controls introduced afterwards, not evidence that the revised process found suitable customers. The post gives no later conversion figures or confirmed revenue estimates. Its value is the specific boundary between producing an outreach batch and deciding whether those people should receive it.

Prompt
Suggested prompt (written by Agents In Use, not the author's)

Source: X · @malmaladei

Frequently asked questions

Why were so many leads the wrong size?

Hoffmann says the bot relied on brand visibility as a revenue signal, which favoured larger, more visible businesses.

What changed after the failed qualification pass?

He required approval before sending, added explicit disqualifiers and told the bot to mark uncertain signals unclear.

Did the revised workflow produce qualified customers?

The capture gives no follow-up results or conversions. It documents the initial mismatch and the controls introduced afterwards.

More details

Remaining steps

  1. He added batch approval before sending, explicit disqualifiers and an unclear flag for uncertain signals.

About this account

Grok Bot researched and scored prospects and drafted personalised outreach. Under the revised rules it flags uncertainty and presents each batch for approval rather than sending autonomously.

Source: verbatim X public API capture supplied by Scout, who read the post live on 2026-10-31. Entry reviewed against that supplied text, not by reopening X. Outcomes are poster-reported, not independently reproduced. Attached media and self-thread continuations were not reviewed. Prompt: suggested by Agents In Use, not a quoted original; any extra approval or checking steps in it are editorial suggestions. The author owns getease; included for his candid Grok Bot qualification failure and approval controls, not as a recommendation of his startup.

Credit: @malmaladei

Source published

Added to Agents In Use

Source checked

Tools and topics

Grok Bot

Lead qualification · Shopify · Outreach · Human approval

Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.