This is an n8n workflow our team at Ryz Labs uses for sales outreach. Every day it researches Russell 2000 companies, finds the CEO's email, writes that CEO a customized email with ChatGPT, logs it all to Google Sheets and sends the email from Gmail. Jordan walked through it and ran it live in Episode 17 (October 17, 2025), starting at 12:14. We didn't say how long it took to build.
Plain and simple: to automate sales outreach. Researching a company, finding the right contact and writing a tailored first email is slow when a person does it, and it is the same steps every time. That is what n8n is for. Jordan describes it as a platform for agentic workflows, not that different from Zapier or Make, that connects to most major tools and to third-party APIs like OpenAI and Perplexity.
Jordan summed up the flow at 13:08:
So basically, what this n eight n automation workflow does is it researches Russell 2,000 companies.
— Jordan Metzner, Episode 17
The manual run worked on air. You could watch the logs move through each node, from getting companies to research to the ChatGPT calls, and n8n showed how many tokens each OpenAI call used. A full run takes a few minutes, so we didn't wait for the last email to go out on camera.
Sam's point about n8n in general: it is "theoretically low code or no code," but there is a learning curve and it is daunting at first. Our way around that is to have Claude build about 90% of an automation, then tie it together at the end. Right after this demo, Jordan did exactly that live, which became the Claude-generated n8n podcast guest outreach workflow.
Jordan also compared it with OpenAI's new agent workflow tool, which he had tried and called "very, very rudimentary" next to n8n.
The episode didn't cover reply rates or results from the emails, so we can't tell you how well the outreach converted. If you build one, expect the last 10% to be the hard part: connecting accounts and APIs is where Sam said things "can be at times challenging."
The same episode also covered our Sora 2 API video generator. For more on AI in outbound, see AI in Sales.
Our workflow pulls a list of Russell 2000 companies, sets research questions, researches each company with Perplexity, gets the CEO's email from a contact-data API, has ChatGPT write a structured custom email, logs everything to Google Sheets, then saves and sends the draft from Gmail.
Yes. In our workflow Perplexity does the company research and ChatGPT, through OpenAI's APIs, turns that research into structured output and the final email. n8n shows token usage for each OpenAI call.
We added wait steps between calls so the workflow does not hammer any API. It runs on a daily schedule and can also be executed by hand.
There is a learning curve even though it is low-code. Our shortcut is to have Claude generate about 90% of a workflow as JSON, import it, and then do the last 10% of connecting accounts by hand.
