Lead Enrichment Pipeline - Deepline

Human guide: work with your agent →

Build a small lead-enrichment pilot with separate company, person, and validation outcomes.

Use this recipe when you already have raw leads and want fast, structured company + person context.

Workflow

Seed leads from a CSV.
Normalize company data first.
Enrich person/role fields with a second pass.
Add quality gates (domain/email validation) before downstream automation.

Keep each block single-purpose so failures are obvious and reruns are cheap.

Claude Code Prompt

I have leads.csv with person_first_name, person_last_name, and company_name.
Run a lead enrichment pipeline with this order:
1) company normalization
2) people enrichment
3) email validation

Start with rows 0-20 only and write output to leads-enriched.csv.
Before running, show the exact action plan and expected output columns.
After running, summarize:
- success/failure count by action
- 3 representative enriched rows
- any schema or quality risks before scaling

Why This Layout Is Reliable

Keeps each step single-purpose.
Allows quick pilot runs before running the full file.
Makes failures easy to diagnose because each block maps to a clear provider.

View source on GitHub | I Have X, I Want Y →

Verify this task

Reconcile input IDs and each stage’s output. Keep provider errors separate from missing or rejected fields. Inspect actual scoped usage. Ask your agent to explain a failed check before repeating work or expanding the scope.