I stopped reviewing Gold leads. I gave them to an AI agent instead.

I'm Jeremy Hurst, VP of Autonomous GTM at Swan. I left a $450K VP Sales role to take a job with a made-up title. I did it because I got a glimpse of the 100x seller - it stopped being a thought experiment and became the actual playbook. This is a field journal for GTM and growth leaders who can feel the model changing under their feet and prefer changing with it to being left behind.

We're generating $1M+ of pipeline per month with 1 GTM person (me) + AI agents. No paid advertising. No SDRs. And I'll be honest... I'm struggling to keep up.

For context, 2 months ago I walked away from a $450K job to become a VP who manages agents instead of people.

To make this work, lead scoring is everything. It's how I know where to spend time vs. where to let the agents run.

Here's how we do it:

💎 Diamond - highest intent accounts with highest ACV ceiling. I own these personally. Every touchpoint.

🟡 Gold - strong first-party intent, real deal potential. I'm in the loop on every sequence.

⚪ Silver - some signals, but not enough intent or ACV to justify a human in the loop. Agents handle everything.

🟤 Bronze - floor tier. Watching & waiting.

But there's a problem. MQLs are piling up faster than I can process them.

Manually reviewing every sequence once felt like the right thing, but it's now feeling like a bottleneck.

So I'm making a call.

Gold leads go fully autonomous, effective tomorrow.

An agent researches the account, finds the people, builds the sequences, and executes. No human review.

It's a scary proposition - I won't pretend otherwise. But it's the only way to scale.

Here's why I'm comfortable doing it.

At this point, the outreach agent gets it right 70-80% of the time. I can live with that failure rate.

If 70% quality means processing 2 or 3x more leads, the net is massively positive. I'd rather handle the error rate than cap the volume.

Think about what we expect from new sales hires. Weeks, months, even a year in ENT before they're executing end-to-end. Every rep gets a ramp. I'm giving the agent the same grace.

That said, I'm not going in blind.

I've got time blocked Tuesday and Friday next week to read through everything the agent sent out. Looking for messaging drift, glaring errors, anything off.

If it looks clean, I'll drop to once a week. Then monthly. Then quarterly.

I'm slow rolling it because MQL-to-meeting booked is one of the most important conversion rates we have. Can't afford to play fast & loose.

Here's the thing about AI.

I won't ever know where the boundary is if I don't push the limit. I'd rather test what's possible than default to what worked when scale meant headcount.

Will be sure to report back with results.

getswan.com

THE SIGNAL

$1M+ pipeline, 1 person, no SDRs. The bottleneck isn't the agents. It's the human in the loop.

THE PLAY

Figure out where the human is the bottleneck before the volume figures it out for you.

• Map your leads into tiers by ACV ceiling and intent signals. If you don't have tiers, that's step one. You can't automate what you haven't defined.
• Identify which tier is creating the review bottleneck. Where are MQLs piling up? That's where you're losing volume to process.
• Run the math. What's your agent's current quality rate? At 70%, 3x volume means the net is massively positive even with errors.
• Pick one tier to go fully autonomous for 30 days. Start where the failure cost is lowest.
• Block two audit sessions in week one. Look for messaging drift, anything off. If it looks clean, drop to weekly.

SIT WITH THIS

If your agent is getting it right 70% of the time, what are you actually protecting by reviewing every sequence? Whatever the answer is, it's costing you volume you'll never get back.

-Jeremy

[Read more.

](https://www.linkedin.com/posts/jeremyhurstprofile_were-generating-1m-of-pipeline-per-month-activity-7468391579850833920-9Ksu?utm_source=share&utm_medium=member_desktop&rcm=ACoAAANrQt8BPYZxd0AHy0VB-G4pIHAm1IRs3mI)