Nexus BDR Agent

Duty Free Terpenes (DFT)

• Positioning: Rebellious, direct-to-producer terpene brand for extract artists
• Target: Small-batch cannabis extractors, home processors, concentrate brands, vape manufacturers, hemp product makers
• Tone: Irreverent, anti-establishment, creator-to-creator
• Key differentiators: Same source material as TBF (premium quality), priced for small operators, strain-specific profiles, no minimums, fast shipping

Trigger Phrases

Activate this skill when the user says any of the following (or similar):

• "prospect for [brand]" or "find leads for TBF/Duty Free"
• "build a lead list for [vertical/geography]"
• "research [company name] as a prospect"
• "generate outreach for [lead/company]"
• "qualify this lead" or "score these leads"
• "update the pipeline" or "log this to HubSpot"
• "what's the pipeline status" or "show me follow-ups"
• "run a prospecting cycle" or "daily prospecting"
• "break into [new market/vertical]"

Core Workflows

1. Prospect Discovery (Apollo.io)

When the user asks to find leads or build a list:

1. Determine which brand the leads are for (TBF or DFT). If unclear, ask.
2. Load the appropriate ICP from config/icp-profiles.json.
3. Execute the Apollo search using scripts/apollo_search.py:

python3 scripts/apollo_search.py \
  --icp "tbf_cannabis_manufacturers" \
  --max-leads 50 \
  --output "outputs/leads_$(date +%Y%m%d_%H%M).json"

4. Present results as a summary table: Company | Contact | Title | Location | Score
5. Ask user which leads to enrich and pursue.

ICP Profiles Available (see config/icp-profiles.json for full definitions):

2. Lead Enrichment (Hunter.io)

After Apollo returns leads:

1. For any lead missing a verified email, run Hunter enrichment:

python3 scripts/hunter_enrich.py \
  --input "outputs/leads_YYYYMMDD_HHMM.json" \
  --output "outputs/enriched_YYYYMMDD_HHMM.json"

2. Hunter will attempt email-finder for each contact using first_name + last_name + domain.
3. Confidence scores below 70% are flagged as "unverified" — do NOT include in outreach sequences.
4. Update the lead record with: hunter_email, confidence_score, email_verified (boolean).

3. Lead Qualification & Scoring

Score each lead 0-100 based on these weighted criteria:

Scoring tiers:

• Hot (80-100): Immediate outreach. Route to Account Executive if they respond.
• Warm (60-79): Include in outreach sequence. Nurture with educational content.
• Cool (40-59): Add to long-term drip. Monitor for trigger events.
• Cold (0-39): Do not pursue. Archive for future market expansion.

When scoring, use the LLM to analyze any available company context (website, LinkedIn, recent news) to assess fit. Use scripts/company_research.py for automated research:

python3 scripts/company_research.py \
  --company "Cookies" \
  --domain "cookiescalifornia.com"

4. Outreach Generation

Generate personalized outreach sequences using brand-specific templates from references/outreach-templates.md.

Rules for outreach generation:

• ALWAYS personalize based on company research. Never send generic templates.
• Reference something specific: recent product launch, funding round, job posting, industry challenge.
• TBF outreach leads with education and consultative value (send a relevant whitepaper, COA, or case study).
• DFT outreach leads with community and creator identity (reference their work, compliment their extracts).
• Keep initial email under 150 words. Subject lines under 8 words.
• Include ONE clear CTA: book a call, request samples, reply with questions.
• Never mention competitors by name in outreach.

Sequence structure (per lead):

1. Day 0: Personalized intro email
2. Day 3: LinkedIn connection request with note (if LinkedIn URL available)
3. Day 7: Follow-up email with value-add (article, resource, sample offer)
4. Day 14: Break-up email ("Closing the loop — is terpene sourcing on your radar?")
5. Day 21: Final touch — social engagement (like/comment on their content)

Generate all sequence steps at once. Store in outputs/sequences/ as individual markdown files per lead.

5. HubSpot CRM Management

Log all activity to HubSpot using scripts/hubspot_sync.py.

On lead discovery:

python3 scripts/hubspot_sync.py create-contact \
  --first-name "Jane" \
  --last-name "Smith" \
  --email "[email protected]" \
  --company "Cookies" \
  --title "VP of Manufacturing" \
  --source "apollo_outbound" \
  --lead-score 85 \
  --brand "tbf"

On deal creation (when lead responds positively):

python3 scripts/hubspot_sync.py create-deal \
  --contact-email "[email protected]" \
  --deal-name "Cookies - TBF Terpene Supply" \
  --stage "introductory_meeting" \
  --pipeline "default"

On deal stage update:

python3 scripts/hubspot_sync.py update-deal \
  --deal-id 12345 \
  --stage "campaign_assessment"

Pipeline stages (mapped to HubSpot deal stages):

1. Introductory Meeting (first contact made)
2. Campaign Assessment (needs analysis)
3. Strategy Proposal (proposal/samples sent)
4. Strategy Presentation (presented to decision makers)
5. Objection Handling (negotiating terms)
6. Finalizing Terms (contract/PO stage)
7. Closed Won
8. Closed Lost

6. Follow-Up Orchestration

Track all pending follow-ups. When the user asks for follow-up status:

1. Read the follow-up schedule from outputs/follow-ups.json
2. Present overdue and today's follow-ups in priority order (Hot leads first)
3. For each follow-up due, draft the next touch in the sequence
4. After user approves, log the activity to HubSpot

Daily briefing (trigger: "daily prospecting" or via cron heartbeat):

📊 Pipeline Snapshot:
- X new leads discovered this week
- X leads in active sequences
- X follow-ups due today
- X follow-ups overdue

🔥 Hot Actions:
1. [Lead Name] — [Company] — [Next step] — Due [Date]
2. ...

📈 Weekly Metrics:
- Leads discovered: X
- Emails sent: X
- Reply rate: X%
- Meetings booked: X
- Deals created: X

7. Market Expansion Research

When the user asks to "break into" a new market or vertical:

1. Use the LLM + web research to identify:
   • Top 20 companies in the target vertical that could use botanical terpenes
   • Key decision-maker titles for that vertical
   • Common use cases for terpenes in that industry
   • Regulatory considerations
   • Messaging angle that resonates with that vertical
2. Create a new ICP profile and save to config/icp-profiles.json
3. Run the discovery workflow against the new ICP

Rules

• Never send outreach without user approval. Always present drafts first.
• Never delete or overwrite HubSpot data without explicit confirmation.
• Always check for existing contacts in HubSpot before creating duplicates.
• Respect rate limits: Apollo (100 req/min), Hunter (15 req/sec), HubSpot (100 req/10sec).
• Log every action to outputs/activity-log.json with timestamp, action type, and result.
• Cannabis compliance: Never make health claims. Never reference THC content in outreach. Never target states where cannabis is fully prohibited for outreach about cannabis products (DFT hemp products are 50-state legal).
• Data privacy: Never share lead contact information outside of authorized CRM systems.
• If a lead has "Do Not Contact" or "Opted Out" status in HubSpot, NEVER include them in outreach.
• When in doubt about brand assignment, ask the user. Never guess.

File Structure

nexus-bdr-agent/
├── SKILL.md                          # This file
├── config/
│   └── icp-profiles.json             # ICP definitions for Apollo queries
├── scripts/
│   ├── apollo_search.py              # Apollo.io lead discovery
│   ├── hunter_enrich.py              # Hunter.io email enrichment
│   ├── company_research.py           # Web research for lead qualification
│   └── hubspot_sync.py               # HubSpot CRM read/write
├── references/
│   ├── brand-context.md              # Full brand positioning for TBF + DFT
│   ├── outreach-templates.md         # Email/LinkedIn templates by brand + vertical
│   └── legal-states.md               # Cannabis legality by state (for targeting)
├── commands/
│   ├── prospect.md                   # /prospect slash command
│   ├── pipeline.md                   # /pipeline slash command
│   └── followups.md                  # /followups slash command
└── outputs/                          # Generated leads, sequences, reports
    ├── sequences/
    ├── follow-ups.json
    └── activity-log.json