LinkedIn B2B Pre-Nurturing Engine

Context

Cold B2B outreach in Latin America has extremely low conversion rates. Decision makers need 7-13 touchpoints before considering a meeting, but monitoring 50-100 prospects manually takes 30+ min/day and is the first thing dropped when schedules get busy.

Solution

A two-workflow pipeline that eliminates the cognitive heavy lifting: Workflow 1 detects new posts every 6 hours and generates strategic comment drafts via Claude; Workflow 2 publishes approved comments via LinkedIn API. A human reviews and approves — protecting account credibility while reducing daily effort by 70%.

Results

• Daily effort reduced from 30-45 min to ~10 min (review only)
• Prospects monitored: 10-15 manually → 50-100 simultaneously
• Touchpoints per prospect/month: 2-4 irregular → 8-12 systematic
• Consistency guaranteed regardless of daily workload
• Zero generic comments — each draft is context-aware and strategic

Architecture

WORKFLOW 1 — Detection & Drafting (every 6h)
Schedule → NocoDB (prospects + history)
  → SplitInBatches (1 contact per iteration)
    → Apify (linkedin-post-search)
    → Code: filter new posts (Set lookup O(1))
    → IF new post? → Claude API → draft → NocoDB (Pending)
    → No new post? → next contact

WORKFLOW 2 — Controlled Publishing (every 5 min)
Schedule → LinkedIn API (get profile URN)
  → NocoDB (filter Approved comments)
  → SplitInBatches
    → Code: normalize URN
    → LinkedIn Social Actions API (POST comment)
    → NocoDB (update: Published)