AI Automation · Outbound Systems · 2024

LinkedIn AI Automation: 6.7x SDR Throughput

System: n8n + Claude API + LinkedIn  ·  Use Case: AI-Powered B2B Outreach at Scale
6.7x
Throughput vs Human SDR
6.7x
SDR Throughput
AI
Personalisation at Scale
Throttled
Safe Send Rates
n8n
Automation Stack

01 — Context

LinkedIn outreach is the dominant B2B prospecting channel. It is also the most manually intensive — researching prospects, writing personalised messages, managing follow-up sequences, and staying within LinkedIn's rate limits to avoid account restrictions.

This system was built to automate the entire outreach workflow end-to-end, from lead enrichment to personalised message delivery, while maintaining the quality of a hand-crafted outreach sequence. The architecture became the foundation for the Cybersecurity AI SDR and the Arkh Media outreach service model.

02 — The Challenge

Personalisation vs. Scale

Generic outreach at scale is spam. Genuinely personalised outreach at scale is humanly impossible — a human SDR can research and write to 30–50 prospects per day with quality. The challenge was building a system that could process hundreds of prospects per day while producing outreach that felt individually crafted, not templated.

Platform Safety

LinkedIn actively detects and restricts automated behaviour. Sending too many requests, moving too fast, or following patterns inconsistent with human behaviour triggers account restrictions. Any automation system must mimic human-like timing and respect daily volume limits.

03 — Strategy & Technology

Lead Import (CSV/D7)
→
Enrichment
→
AI Personalisation
→
Throttled Send
→
Follow-Up Sequence

The n8n Automation Workflow

01. IMPORT — Leads ingested from CSV export or D7 Lead Finder
02. ENRICH — Domain scraped for company context, job title, recent activity
03. SCORE — Claude API scores lead relevance against ICP criteria
04. DRAFT — Claude generates personalised connection request + follow-up sequence
05. THROTTLE — Requests batched with human-like timing delays (random 2–8 min gaps)
06. SEND — Unipile API executes LinkedIn actions within daily safe limits
07. SEQUENCE — Automated follow-up triggers on connection accept + no-reply logic

AI Personalisation Layer

Claude API generates outreach based on three inputs: the prospect's LinkedIn profile content, their company's recent activity or positioning, and the specific ICP angle being targeted. Output is a connection request note (under 300 characters) and a 3-step follow-up sequence, each message using different angles — not the same message on repeat.

Throttling & Safety Logic

Daily connection request limits enforced at the workflow level. Randomised send timing to avoid detection patterns. Automatic pause triggers if acceptance rate drops below threshold — a signal that messaging is off or account is flagged. Designed to run indefinitely without manual intervention.

04 — Key Metrics

6.7x
Throughput vs Human SDR
100%
Personalised Messages
0
Manual Steps Required
3-step
Automated Follow-Up

What They Said

"Sankalp has worked in my team for four years. His problem solving skills are top notch. He treats every client differently and innovates based on the situation. Sankalp has been an exceptional manager and has been able to get his team to perform at their full potential. He gives them space yet makes them deliver. Sankalp stands out as a person who takes risks and confronts problems instead of avoiding them. This quality makes him a strong manager and professional."
Sharang Arvind Director · Strategy · Sales CoE · PMO · Sodexo India
Automation
n8n
Claude API
LinkedIn
B2B Outreach
AI SDR
← Back to Case Studies