Follow Up Nurture

Follow-Up Nurture — Context-Aware Lead Nudge

Goal

Go through every lead marked "pending" in the database, pull all prior email communications, research anything mentioned that you don't know about, and send a personalized follow-up email in Nick's voice. This isn't a cute demo — it makes money.

Inputs

• Lead database: data/leads_database.json — contains leads with status, email history, notes
• Tone: Casual, direct, helpful. Never salesy. Never use exclamation marks. Sound like a smart friend who happens to run an agency.
• Mode: Always present output as final emails being sent (not "drafts"). Use --generate then --review to produce and display them.

Process

Step 1: Load Pending Leads

python3 execution/follow_up_nurture.py --list-pending

This reads data/leads_database.json and shows all leads with status: "pending".

Step 2: For Each Pending Lead

1. Pull email history — read the email_history array for that lead
2. Identify context gaps — if the lead mentioned something specific (a project, a tool, a problem, a competitor), and you don't have enough context to write intelligently about it, do a quick web search
3. Write follow-up — write a short, context-aware email that:
   • References something specific from your last conversation
   • Adds value (a relevant insight, resource, or observation)
   • Has a soft CTA (no "schedule a call" unless they asked)
   • Is 3-5 sentences max

Step 3: Generate Emails

python3 execution/follow_up_nurture.py --generate

This processes all pending leads, generates emails, and saves to data/followup_emails.json.

Step 4: Review & Send

# Review emails
python3 execution/follow_up_nurture.py --review

# Send emails
python3 execution/follow_up_nurture.py --send

Nick's Tone Guide

Keep it short and low-effort. These are check-ins, not pitches. Match the tone of the original campaign thread.

Templates (pick one based on prior thread tone)

If prior emails were lowercase/casual:

hey {name}, circling back on {topic}. let me know if I can answer any q's?

- nick

If prior emails were title case / slightly formal:

Hey {Name},

Just checking in on {topic}. How are things going? Let me know if you need me.

Thanks,
Nick

If prior emails were warm/friendly:

Hi {Name}—hope you had a great week. Checking in on {topic}. Let me know where we're at.

Thanks,
Nick

If checking in on a general relationship (no specific topic):

Hi {Name}. How are things going? Hope you're crushing it. I'm checking in on {topic}. Let me know if I can help you/answer q's.

Thanks,
Nick

If following up on a sent deliverable (proposal, contract, deck, etc.):

Heyy {Name}. Just saying hi. I know it's been a few days since I sent over that {deliverable}. Let me know if I can answer q's or clarify anything.

Thanks,
Nick

Tone rules

• No "I hope this email finds you well" — ever
• No exclamation marks
• Reference the specific thing you discussed, not a generic "our product"
• 2-3 sentences max. These are nudges, not essays
• Match the formality level of whatever thread you're replying to

Scheduling

This Skill can be run on a schedule (daily, weekly) to automatically nurture your pipeline. The script supports --cron mode for unattended operation.

Output

• data/followup_emails.json — generated emails with lead context
• Sent emails logged back to data/leads_database.json under each lead's email_history

Successive Run Behavior

This skill is designed to run daily. Each follow-up must be different from the last:

• The prompt sees the full email history including prior follow-ups
• If the last outbound email was already a nudge, the model picks a different template and phrasing
• This prevents sending "circling back on the proposal" 7 days in a row

Edge Cases

• No prior emails: Skip the lead, flag as "needs intro email instead"
• Lead replied recently (< 48h): Skip, they're already engaged
• Lead marked "closed" or "lost": Skip entirely
• Research fails: Write follow-up without the research, note the gap

First-Run Setup

Before executing, check if the workspace has a .gitignore file. If it doesn't, assume the user is new to this skill. In that case:

1. Ask the user if this is their first time running this skill
2. If yes, walk them through how it works and what they need to configure/set up (API keys, env vars, dependencies, etc.)
3. Let them know that Nick wishes them the best!