We had a campaign we wanted to get in front of the right people. The problem was familiar: We had a targeted list of business leaders we genuinely thought would benefit from what we were promoting, but no clean process for actually reaching them at scale. And we didn’t have enough time to do it the slow way [read: without AI].
So instead of defaulting to the usual approach, I decided to try something different. I wanted to see what an agentic AI system, Claude Code, could do with a cold outreach project from end to end. Not as a production system, but as an experiment. But more importantly, as a way to challenge our team to think differently about how we approach work like this.
What the Old Approach Looks Like
Most marketers know the traditional cold outreach process: get a list, research each prospect, write a template email, customize where you can, then spend a few hours copying, pasting, and hitting send. It works, but it’s slow. And honestly, it’s often the kind of work that doesn’t get done at all because there’s always something more urgent to finish.
A New Approach with AI
I started by directing Claude Code to a webpage that described what we were promoting. From there, I had it work out who the ideal audience would be, including the kinds of roles, seniority levels, and company types we’d want to reach. It mapped that out on its own, based entirely on what it read.
Then, mostly out of curiosity, I let it try to find actual prospects. It identified people it thought were good fits and made educated guesses about their email addresses based on research into common email formats at those companies. I want to be clear: I’m not saying this is the right way to build a verified prospect list. A tool like Clay is the better path for that. But watching it reason through the approach on its own was genuinely interesting.
The part that really mattered came next
Working together, we developed a strong outreach email with a message that was actually relevant to the recipient. Once that was dialed in, I connected Claude Code to my personal email inbox for the demonstration, and it automatically generated 250 draft emails, each personalized with the recipient’s name and relevant details.
It then created an HTML-based “email hub” with each recipient and a button to “Send Email,” which automatically opened the full personalized email in Gmail with each email address populated.
All I had to do was jump into a single browser window, click each entry, and hit send. It took about 20 minutes all-in to complete my outreach.
What This Actually Means for Marketers
A few things stood out from this experiment that I think are worth noting.
- First, none of this replaces the human judgment required for outreach to work. Spammy, irrelevant outreach fails no matter who or what sends it. The quality of the email and the relevance of the audience still have to be determined by you.
- Second, the value here isn’t speed alone. It’s that work like this actually gets done. Our team might have approached this the traditional way and run out of time before it ever launched. An AI-native approach makes previously impractical work practical.
- Third, this is just the beginning of what’s possible when you connect AI agents to the systems you already use. Email is an obvious starting point, but the same logic applies across CRM workflows, follow-up sequences, and campaign personalization at scale.
The tools are here. I recommend experimenting with them before you actually need them or you’ll wind up scrambling to catch up.
Want to learn more about applying AI agents to your workflows? Join us at our next Intro to AI virtual event or our virtual B2B Marketers Summit on June 25, 2026.
