Run these after completing the research step. These use the segment and role data to write sequences that are specific to each lead group.

Prompt: Write LinkedIn DM sequences per segment

Read scored-leads.csv, segment-research.md, and icp-filter.md.

For each segment group (by title + industry), write a 3-message
LinkedIn DM sequence.

Rules:
- Message 1: Under 75 words. Lead with their reality, not your product.
  No pitch. End with a soft question or observation.
- Message 2 (Day 3): Add a data point, case study, or insight relevant
  to their segment. Still no hard ask.
- Message 3 (Day 7): One clear CTA. Specific, not 'open to a call?'
  Frame it around what they get, not what you want.

Do not use: 'Just checking in', 'Quick question', 'Hope you're well',
'I came across your profile', or 'I'd love to connect'.

Output a file called dm-sequences.md with one section per segment.

Prompt: Write cold email sequences per segment

Read scored-leads.csv, segment-research.md, and icp-filter.md.

For each segment group, write a 3-email cold email sequence.

Rules:
- Subject lines: Under 6 words. No clickbait. No questions as subject lines.
- Email 1: Under 120 words. Problem-first. One CTA at the end.
- Email 2 (Day 4): A different angle or proof point. Not a re-pitch.
- Email 3 (Day 10): Short break-up email. Under 50 words.
  Give them a low-friction out.

Format per email: Subject | Body | CTA

Output a file called email-sequences.md with one section per segment.

Prompt: Personalise messages for Tier 1 leads

Run this after the segment sequences are written. It takes the segment template and adds individual context from the tier1-research.md file.

Read dm-sequences.md, email-sequences.md, and tier1-research.md.

For each Tier 1 lead:
1. Find their segment sequence in dm-sequences.md
2. Add one personalised line at the start of Message 1 that references
   the specific signal found in tier1-research.md
3. Keep the rest of the message the same

The personalised line should:
- Be under 20 words
- Reference something real (the post, news item, or signal found)
- Not compliment them on it. Just reference it as context.

Output a file called tier1-personalised-messages.md.
Format: Lead name | Personalised line | Full message

Prompt: Write connection request notes

Read scored-leads.csv and tier1-research.md.

For Tier 1 leads only, write a connection request note.

Rules:
- Under 200 characters (LinkedIn limit)
- No pitch
- Reference something specific to them or their company
- Do not mention you want to 'connect' or 'chat'
- Do not use 'I came across your profile'

For Tier 2 leads: leave blank.
Blank connection requests outperform generic notes for this tier.

Output results in tier1-personalised-messages.md under a new column:
Connection_Note