AI-Powered Outreach Automation for a B2B Logistics-Tech Company
Turning cold outreach from a manual, full-time job into a self-running system — end to end, from lead to reply.
30-day live snapshot.
Fully autonomous over a 30-day live period — no manual sending, writing, or monitoring. Client identity withheld under a confidentiality agreement; all figures are unaltered measured results from the live system.
Personalised outreach converts but doesn’t scale. Automated outreach scales but doesn’t convert.
The client, a B2B software company selling AI-powered dispatch and fleet-control tools into the DACH logistics market (Germany, Austria, Switzerland), needed to reach a large, specific list of logistics companies with outreach that felt personal — not templated. In this industry, generic mail-merge emails are an easy way to get ignored, or worse, marked as spam.
The problem was the usual tradeoff:
- Personalised outreach converts, but doesn’t scale. Researching a company, understanding what they do, and writing a genuinely relevant email takes real time per lead — realistically 10–20 minutes for one good cold email. At any meaningful volume, that becomes a full-time job for an entire team.
- Automated outreach scales, but doesn’t convert. Traditional mail-merge tools fill in a name and company and call it personalisation. Recipients can tell.
On top of that, the client had no simple way to see what was happening. Sending was split across multiple inboxes to stay within safe daily limits, replies came back to different accounts, and there was no single place to check what had gone out, what had bounced, or who had responded — without digging through spreadsheets and inboxes one by one.
They needed a system that could do the research, write like a person, send safely at scale, and report on itself — without anyone babysitting it.
A complete outreach automation platform.
A network of coordinated workflows built in n8n that takes a lead from a raw name on a spreadsheet to a fully researched, AI-written, safely delivered email — plus a custom live dashboard giving the team full visibility in real time.
- Researches every lead automatically — pulling their LinkedIn profile and scraping their company website for real, relevant context, with automatic fallbacks if a scrape fails, so no lead silently drops out of the pipeline.
- Writes a genuinely personalised email per lead — using Claude, grounded in that specific research — not a mail-merge template with blanks filled in.
- Sends safely at scale — by rotating delivery across 15 separate inboxes with capped daily volumes and randomised human-like delays, so sending stays within safe limits and protects deliverability.
- Tracks everything automatically — opens, replies, and follow-up sequences, with zero manual monitoring.
- Reports live — to a custom dashboard and an instant ops-alert channel, so any failure is caught within minutes rather than discovered days later.
Step by step.
Automation is only useful if people can see what it’s doing.
Alongside the pipeline, we built a live operations dashboard that turns every workflow event into something the team can act on.
Overview
Open rate, reply rate, volume, and the latest replies, at a glance.
Emails Sent
A full searchable log of every send, filterable by inbox, sequence step, and language.
Replies
Every lead conversation, organised by inbox, in a clean thread view.
Alerts
Every failed send or failed AI generation, with full context, the moment it happens.
Volume & Accounts
Sending health across all 15 inboxes, so no account is ever at risk of being overused.
Templates & AI Script
The follow-up messaging and the AI’s writing instructions, editable directly by the team, with no engineering involvement required.
Nothing on this dashboard is a static report. Every screen updates live, straight from the automation itself.
Time reclaimed.
Manually researching and writing one well-personalised cold email realistically takes a skilled SDR 10–20 minutes. Applied to the 4,137 emails the system produced in 30 days, that same volume would have required:
| Manual pace assumption | Estimated hours for 4,137 emails | Equivalent full-time effort |
| 10 minutes / email | ≈ 690 hours | ~1 person, full-time, for over 4 months |
| 15 minutes / email | ≈ 1,034 hours | ~1 person, full-time, for over 6 months |
| 20 minutes / email | ≈ 1,379 hours | ~1 person, full-time, for over 8 months |
Based on a 40-hour work week. Actual per-email research time varies by SDR experience and lead complexity. To translate this into a dollar figure, multiply the hours above by your organisation’s fully-loaded hourly cost for an SDR — base pay, benefits, tools and management overhead.
Beyond the send numbers.
- No dedicated outreach team required. What would normally require several full-time sales development reps — researching leads, writing emails, managing multiple inboxes, tracking replies, and following up — now runs unattended, with the client’s team only stepping in to review replies and refine messaging when they want to.
- Deliverability protected, not just volume achieved. By automatically capping and rotating sends across 15 inboxes with human-like pacing, the system hit high volume without putting sender reputation at risk — something that’s genuinely difficult to manage manually at this scale.
- Nothing falls through the cracks. Every failure — a bad lead, a failed AI generation, a bounced send — is caught and surfaced immediately instead of being discovered later as a missed opportunity.
- Full visibility without technical overhead. The client’s team can monitor performance, adjust messaging, and manage sending accounts entirely from the dashboard — no access to the automation platform or engineering support needed for day-to-day operation.
What it’s built on.
| Orchestration | n8n |
| AI / copywriting | Claude (Anthropic) |
| Lead research | Automated LinkedIn & website scraping via Oxylabs Studio and BrightData, with fallback extraction |
| Lead source of truth | Google Sheets |
| Sending | Gmail API, rotated across 15 accounts |
| Ops alerting | Real-time Telegram alerts |
| Dashboard | Custom-built live web application |
| Tracking | Custom open-tracking and reply-capture system |
Measured over a 30-day live period, entirely hands-off.
- 4,137 emails sent autonomously
- 50.1% open rate
- A 3.5–4.8% reply rate, well above typical B2B cold-outreach benchmarks
- 15/15 sending accounts kept in healthy standing throughout
- 4,137 leads processed across 32 completed batches
- An estimated 690–1,379 hours of manual research and writing work eliminated
- Zero manual monitoring required to keep the system running
This project wasn’t about building one automation — it was about designing a small, self-monitoring outreach operation: research, AI writing, multi-account sending, reply and open tracking, follow-ups, error recovery, and live reporting, all working together and visible from a single screen. The result is a system the client’s team can run and refine on their own, without needing to understand — or even see — the automation running underneath it.
Client name withheld under a confidentiality agreement. All performance figures on this page are real, unaltered results from a live 30-day deployment.