Key takeaways
- We stopped hiring SDRs for top-of-funnel prospecting and built an n8n + Claude agent to do it instead.
- The agent researches each lead, qualifies against our ICP, and drafts a personalized first email — a human only approves before send.
- It runs as four connected n8n workflows — sourcing, research and drafting, human approval, and campaign push with follow-up — not one giant script.
- The system is built to sustain real outbound volume without adding headcount, with a human-in-the-loop gate before anything reaches a prospect's inbox.
- The same architecture — research, qualify, draft, escalate to a human, execute — applies to any repetitive research-and-outreach job, not just sales.
The Job We Stopped Hiring For
Every SDR hire follows the same script: find companies that look like your best customers, figure out who to email, write something that doesn't sound like a template, and follow up three times before giving up. It is a research job wearing a sales job's title. Most of the actual output — the finding, the qualifying, the first draft — has nothing to do with charisma. It's pattern-matching against a target profile and writing a coherent paragraph. That's the part we stopped paying a human to do.
We didn't outsource it to a cheaper research agency, and we didn't buy a tool that promises "AI SDRs" as a marketing line. We built it — an n8n workflow that runs the entire top-of-funnel research and outreach loop, with Claude doing the reading and writing, and a human reviewing before anything goes out.
What Actually Sat on an SDR's Desk
Strip away the job title and a prospecting SDR's week is four repeatable steps: pull a list of companies matching an ideal customer profile, research each one enough to know if they're a real fit, write a first-touch email that references something true and specific about them, and queue the follow-ups. None of that requires judgment that changes case to case — it requires consistency, which is exactly what humans are bad at across a 40-hour week and software is good at continuously.
The actual cost isn't just an SDR's salary. It's the variance: a good week produces sharp, specific emails; a tired Friday produces "Hi {{firstName}}, hope you're well" templates that convert at a fraction of the rate. An agent doesn't have a tired Friday.
The Workflow, Not the Node Diagram
Our pipeline runs as four connected n8n workflows, not one script trying to do everything:
- Sourcing: pulls candidate leads against our ICP and drops them into a queue.
- Research and drafting: for each lead, Claude reads real signals — company activity, role, public context — decides whether the lead actually fits, and drafts a personalized first email. This is the step that used to be a person's whole morning; it now runs unattended.
- Human-in-the-loop gate: nothing reaches a prospect's inbox without passing through a review queue first. The agent proposes; a human approves. This isn't a hedge — it's the actual control point that makes the rest of the system safe to run unattended.
- Campaign push and follow-up: approved emails go out through the sequencing platform, with follow-ups scheduled automatically, and results sync back to a shared pipeline so nothing gets double-touched.
Each stage is a separate, small n8n workflow with one job. That's a deliberate choice: if the research step needs a better prompt, we change one workflow, not the whole system.
What This Actually Replaces
It doesn't replace the person who gets on a call and closes. It replaces the research-and-first-draft labor that used to consume most of an SDR's day before a single reply ever came back. The system is engineered to sustain outbound at a genuine daily volume — the kind of throughput that would otherwise need multiple full-time researchers just to keep the top of the funnel fed — without adding headcount for that specific function. That's the honest claim: not "AI replaces salespeople," but "AI replaces the repetitive research-and-drafting labor that was never actually the sales part of the job."
Why This Isn't Just a Sales Story
The pattern underneath — research a target, decide if it qualifies, draft a first pass, escalate to a human before anything commits — is not sales-specific. It's the same shape as a paralegal screening incoming matters, a bookkeeper classifying transactions before they hit the ledger, or a support rep triaging tickets before they're answered. Once you've built the architecture once, the vertical is a configuration detail, not a rebuild.
Frequently Asked Questions
Does this replace our CRM or email tool?
No — it sits on top of them. The agent researches and drafts; your existing CRM and sending platform still own the record and the send. The workflow is the connective tissue that was previously a person manually operating both.
What stops the agent from sending something wrong?
The human-in-the-loop gate. Every draft is queued for approval before it reaches a real inbox. The agent's job is to get the first draft right often enough that approval is fast, not to remove the approval step.
How long does something like this take to build?
Weeks, not months, for a well-scoped first version — the architecture (source, research/draft, approve, send/sync) is reusable; the specific ICP logic and prompt tuning for your business is the part that takes iteration.
Does this only work for outbound sales?
No. The same four-stage shape — source, research and draft, human approval, execute and sync — is what we reuse across research-heavy, repetitive-decision workflows in other functions too.
