On the Third Company, I Finally Stopped Solving GTM with Headcount

On the first company, I thought the answer was hiring faster.

Get a marketing person. Get a salesperson. Get an agency. The pipeline wasn’t moving — bring in people who could move it. That was the thesis. That’s what growth looked like in every company I’d watched from the outside.

So I hired. People ramped. The pipeline moved — for a while, for as long as those people were engaged and the targeting was fresh. Then something shifted. The salesperson had a bad quarter. The marketing hire moved on. The agency delivered decks but not pipeline. And the growth engine, which had always felt like it ran on people, stalled the moment the people did.

I didn’t learn the lesson. I hired again.

The Second Company: Hire Smarter, Same Result

By the second company I’d seen what bad hiring costs. The ramp time. The management overhead. The 90-day hire who leaves at 91 and takes the institutional knowledge with them. I was not going to repeat those mistakes.

So I hired smarter. More specific job requirements. Longer interview process. Better people — genuinely better, with real track records and strong references.

Same result. Still dependent on people to execute every repeatable task. The outreach ran when they had capacity and stopped when they didn’t. The follow-up cadence held for six weeks and then got dropped when someone got pulled into a product issue. The content calendar was solid until it wasn’t.

The failure mode was identical. The people were better. The cost was higher. The dependency was the same.

What I Was Actually Solving For (And Getting Wrong)

The problem I was trying to solve was: the repeatable execution work of GTM doesn’t happen consistently unless someone is doing it. Research, outreach, follow-up, content, pipeline reviews — all of it requires continuous execution. Not one-time setup. Not strategy. Execution, on a schedule, every week, regardless of what else is happening in the business.

My solution to that problem was always to hire a person (or a team of people) to be the execution layer.

That solution works — until it doesn’t. Until the person gets busy. Until the person leaves. Until the margin math doesn’t support the headcount. Until the team is full and the ceiling appears.

The assumption I never questioned: that the execution layer had to be human. That research, outreach, follow-up, and content required people to run them. That the only way to get consistent execution was to hire consistent people.

It turns out that assumption was wrong, and I was already three companies in before I understood it.

What the Third Company Looked Like Differently

On the third company, I stopped trying to solve the execution problem with better hiring. I separated two types of GTM work that I’d been treating as the same thing:

In every previous company, I’d staffed both categories with humans. Strategic work: people. Execution work: also people, just more of them at lower seniority.

What changed on the third company: the execution layer moved off people entirely and onto an agent system. Sandbox runs the research, the outreach sequences, the follow-up cadences, the content pipeline. I give it direction. It executes. Consistently, on schedule, regardless of what delivery cycles look like.

What Actually Changed in Practice

The shift wasn’t magical. The strategy still requires me. The targeting decisions still require me. The relationship conversations still require me.

What changed:

• Outreach runs every week whether I have time to think about it or not
• Follow-up happens on schedule — not on day 11 when someone finally got to it
• Content goes out consistently, not in bursts between busy delivery periods
• Pipeline reviews surface what needs my attention instead of requiring me to go find it
• The GTM motion runs in the background of everything else instead of competing with it

And critically: the team size stopped being the growth rate limiter. Growth was no longer a function of how many people I could afford to have executing growth tasks. The execution layer scaled without adding headcount.

The Lesson Serial Founders Learn Too Late

If you’ve built more than one company, you already know the ceiling. You’ve hired for GTM and watched the problem move up to the next headcount level. You’ve seen a great salesperson leave and felt the pipeline go with them. You’ve built the ops team and watched management overhead compound faster than revenue.

That experience is an advantage right now — because the mental model shift is much easier when you’ve lived the alternative. “I need more people to execute this” to “I need a system that executes this without people” is not a subtle difference. But it takes a few rounds of the same failure to really want to make it.

The founders moving fastest with Sandbox aren’t the ones excited about AI as a technology. They’re the ones who looked at their second or third company and decided they were done solving the same execution bottleneck the same way.

I wish I’d had it on company one. I’d have saved a lot of ramp time and a lot of severance.