Five Grok Bots outbound: 140 local firms emailed in ~6h, 11 deals self-reported at ~$27k
アイル｜AI事業の作り方 (@isle_ai_biz)X (Twitter) · 2026-10-24
Read originalJapanese X long post that decomposes @ArchiveExplorer's self-reported run: 5 Grok Bots with separate inboxes. Bot1 filters 468 local businesses by 'gap closable in 1 week' (drops 69%). Bot2 logs evidence (stale site, unanswered reviews, 404 booking form). Bot3 builds a one-page screenshot audit. Bot4 sends and handles reply threads from its own inbox. Bot5 kills sends that would annoy the recipient. Result claimed: 140 sent, 27 replies (~19%), 23 call requests, 11 closed; run cost ~$12; revenue so far ~$27,410 at $500–$8,500 per client. Author stresses this is self-reported, not average, and lists Japan-specific legal (特定電子メール法) and security caveats.
Business model
Japanese X account @isle_ai_biz published a long breakdown of @ArchiveExplorer's Grok Bot crew: five agents with separate inboxes and narrow jobs (filter, evidence, one-page audit, send+thread, kill-annoying-sends). Claimed funnel: 468 → 146 → 140 sent → 27 replies → 23 call asks → 11 closed; ~$12 run cost; ~$27,410 revenue so far at $500–$8,500 per client.
Turns lead gen for AI implementation/consulting into a repeatable pipeline that only emails firms with a concrete, one-week-fixable gap and attaches a useful audit—not a mass blast.
Workflow
- Overnight: five Grok Bots, inboxes, charters.
- Bot1 filters on closable gap.
- Bot2 logs screenshot-able evidence.
- Bot3 builds one-page audit.
- Bot4 sends and runs reply threads from its own inbox.
- Bot5 vetoes annoying drafts.
- Human reviews threads as manager and closes.
Adaptation plan
Japanese X long-form summary of an overseas original post. Revenue and close counts are self-reported by @ArchiveExplorer and unverified. Author labels results as first-run, not average, and adds Japan compliance notes. Not a guaranteed income outcome.
Starter prompt
Written by AgentFlow; not the original author’s prompt. Not independently deployed in Muse or Grok.
(Charter pattern, not verbatim) Narrow roles only: Bot1 = closable-in-one-week gap; Bot2 = evidence log; Bot3 = one-page audit with screenshots; Bot4 = send+thread from own inbox; Bot5 = single question would this annoy me. Never touch primary mailbox; no audit = no send.
Running costs and pitfalls
Japanese X long-form summary of an overseas original post. Revenue and close counts are self-reported by @ArchiveExplorer and unverified. Author labels results as first-run, not average, and adds Japan compliance notes. Not a guaranteed income outcome. No independent cost benchmark is available. Check subscriptions, permissions and external actions before running a workflow.