Skip to main content

​
Personalize Outreach

To generate personalized cold emails for a list of contacts, use Deepline’s 3-pass play: research each contact with web search (Claude Code’s native web agent, Exa, or Parallel AI), extract the strongest personalization signals with call_ai, then generate tailored copy in a separate call_ai pass. One natural-language prompt produces emails that reference verified details about each prospect.
Personalized emails outperform generic templates in open and reply rates. Deepline’s 3-pass separation grounds personalization in real research rather than hallucinated details.
The play researches each prospect’s recent activity and company news, then writes a cold email that references something specific to them instead of sounding templated.

​
How do I write personalized cold emails with Claude Code?

Tell Claude Code what contacts to research and what tone you want. The 3-pass approach (research, extract signals, generate copy) avoids the common failure mode of single-pass tools that hallucinate personalization details. Pass a single contact or a CSV of hundreds.
“Research these contacts and write a personalized cold email for each, referencing something specific about their company or role”
“Generate personalized one-liners for contacts.csv based on their LinkedIn activity and company news”
With Codex: 
codex "Research contacts.csv and write a personalized cold email opening for each contact"

​
What does the personalization workflow do step by step?

Research, signal extraction, and copy generation run as three separate passes. Single-pass tools that try to research and write simultaneously tend to produce generic or hallucinated personalization.
1

Research each contact (Pass 1)

For each contact, Deepline gathers context: recent LinkedIn activity, company news, role changes, and public content they’ve authored.
2

Extract personalization signals (Pass 2)

call_ai identifies the strongest personalization hooks: a recent talk, a company milestone, a shared interest, a relevant pain point.
3

Generate personalized copy (Pass 3)

call_ai writes the email or one-liner using the extracted signals, following your tone and structure guidelines.
4

Write output

Results include the personalized email/one-liner, the signal used, and the source for each contact.

​
Which providers power the personalization?

Three tools run across the three passes. The AI writes copy based on verified research, not hallucinated details. 25+ data providers feed richer context than single-provider tools.
  1. Web search — Claude Code’s native web agent, Exa, or Parallel AI gathers recent public information about the contact and their company
  2. call_ai (signal extraction) — Identifies the best personalization hooks from raw research
  3. call_ai (copy generation) — Writes personalized output using the extracted signals
B2B data decays at roughly 22.5% per year (HubSpot via Cognism), so the live web search in Pass 1 ensures your personalization references current events rather than stale data. 
For each contact in contacts.csv:
1. Research their company's recent news and the contact's LinkedIn activity
2. Write a personalized cold email opening (2-3 sentences) that references something specific
3. Keep the tone professional but conversational
4. Write to outreach.csv with columns: name, email, personalized_opening, signal_used
The quality of personalization depends on how much public information exists for each contact. C-level executives at well-known companies produce the best results. For less visible contacts, the play falls back to company-level personalization.
Related: Build Prospect List | Find Work Email | Company Research Brief

​
Frequently Asked Questions

​
How do I personalize cold emails for a large list?

Pass a CSV of contacts to Claude Code: “Research these contacts and write a personalized cold email for each.” Deepline runs a 3-pass workflow — web search for research, call_ai for signal extraction, call_ai for copy generation — producing emails that reference real, verifiable details about each prospect.

​
Why does the 3-pass approach produce better personalization?

Single-pass approaches that try to research and write simultaneously tend to produce generic or hallucinated output. Deepline’s 3-pass separation means the copy-generation step only works with verified signals from the research step. You can inspect the extracted signals before copy generation runs, and adjust the tone or angle by changing the prompt.

​
Can I personalize outreach in bulk from a CSV?

Yes. Point Claude Code at a CSV with contact names, emails, or LinkedIn URLs. Deepline processes each contact through the 3-pass research-extract-generate pipeline. Batches of 100+ contacts work without configuration changes, and output includes the personalized copy, the signal used, and the source URL for verification.

​
What data do I need to run personalized outreach?

At minimum, contact names and company domains. LinkedIn URLs improve results because Deepline can pull the contact’s recent activity and posts. Adding role, company, and LinkedIn URL together yields the most specific personalization.

​
How do I automate personalized outreach?

Combine the personalize outreach play with the Build Prospect List play. First build a list of qualified prospects with verified emails, then run the personalization play against that list. The output CSV can be imported directly into your outreach tool (Outreach, Apollo, Salesloft).