Claude Code queues fifteen researched Freckle emails
Toizer reports a Freckle outbound campaign with 433 shortlisted contacts, 396 enriched emails and fifteen researched Gmail drafts, with Slack reminders and human sending.
Original prompt
Shared by the source author. See source notes for context.
/plan-campaign
Reach people exploring AI agents
See placements →What happened
Toizer reports a Freckle outbound campaign with 433 shortlisted contacts, 396 enriched emails and fifteen researched Gmail drafts, with Slack reminders and human sending.
Reported by Andy ToizerHow the agent did it
- Define the institutional education and construction-software account criteria.
- Combine title searches with identity matching and web fallbacks.
- Test enrichment caching and the fallback before running the shortlist.
Where a person stepped in
Toizer reviews research and copy, narrows the campaign and personally controls sending. Provider costs reflect Freckle’s pricing, not a general price promise.
Andy Toizer, Freckle’s head of growth, uses Claude Code to build an outbound campaign for education and construction software companies. He narrows broad database labels to actual institutional education suppliers and software serving construction businesses, then combines Apollo title searches, AI Ark identity matching and web fallbacks.
The run produces 433 shortlisted prospects and finds 396 emails. Before scaling enrichment, he tests varied contacts, verifies that cached reruns do not spend fresh credits and proves the fallback provider is actually called. These checks make restarts inspectable rather than silently repeating paid lookups.
Writing still needs substantial direction. Toizer rejects stale research, wrong personas and unnatural hooks, and removes marketing and RevOps from the first campaign. Fifteen sales-leader and SDR-manager messages become Gmail drafts. Slack reminders tell him when to review and send, with workday follow-up reminders for LinkedIn.
He remains responsible for sending. Four rate-limit interruptions stretch the initial project across a week, so the result is not an uninterrupted autonomous campaign. The account proves a prepared contact queue and reports subsequent delivery, without publishing campaign conversion metrics. Its useful detail is testing resumability before enrichment and treating human copy feedback as part of execution rather than assuming accurate research automatically produces appropriate outreach.
Prompt
Original prompt
Source: Blog · Andy Toizer
Frequently asked questions
Why test the enrichment cache?
Restarting a failed run should not spend new credits on contacts already enriched.
What reached Gmail?
Fifteen first-batch researched emails became drafts for sales leaders and SDR or BDR managers.
Did it run without interruption?
No. Four rate-limit interruptions broke the flow, and human feedback remained necessary before sending.
More details
Remaining steps
- Revise research hooks and narrow the first recipient personas.
- Create fifteen Gmail drafts and Slack reminders for human review and sending.
About this account
Claude Code selects accounts, integrates providers, scores contacts, runs enrichment checks, researches recipients and prepares Gmail drafts and reminder scheduling.
Firsthand account by Andy Toizer; original publication date verified against Primary Substack byline Apr 17, 2026. Outcomes are the author’s reports, not independently reproduced. Original shared campaign entry command; the publication exposes separate phase skills and keeps final review and sending with the operator.
Credit: Andy Toizer
Source published 17 Apr 2026
Source checked 14 Oct 2026
Tools and topics
Claude Code · Apollo · AI Ark · Lead Magic · Findymail · Gmail · Slack
Enrichment cache tests · Freckle outreach drafts · Human send reminders
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.