Building Under Constraint: The $10M/Employee Playbook
What happens when your team isn’t allowed to grow — and why that might be good.
DEAR STAGE 2: It feels like new startups have an advantage today and can build their processes, systems, and teams around AI, but we’re already a team of 50 and have to “retrofit.” I’m looking to model what this looks like as the CEO. What are small, autonomous teams doing that we can learn from and apply today? ~PLAYING CATCH-UP ON AI
DEAR PLAYING CATCH-UP ON AI: Not the first time I’ve heard this question recently. It’s coming up in board meetings, 1:1s and convos with other investors. Founders are looking around at these agile, AI-native teams and wondering how to rewire what’s already been built.
Last week, I was on a call with a seed-stage founder whose CTO wanted to hire more engineers. Their business is growing fast — the default instinct is to scale headcount. In another board meeting, we were reviewing ballooning support tickets. The suggestion wasn’t to hire more reps. Instead, a board member asked, “Who’s your best problem solver? Give her the challenge: handle 10x the ticket volume without adding headcount.”
If your team solves today's problems with more humans, that will become your default approach — and it’s really hard to reverse. What if, instead, you put an artificial constraint in place?
To explore this idea, I called on Amos Bar-Joseph, CEO of Swan, who’s building a $30M ARR business with just three co-founders. No marketing team, no SDRs, no traditional org chart. Just a radical constraint: $10M ARR per employee. Here’s how Amos is building Swan differently — and how you can apply pieces of his model to your own company.
1. Set an ambitious constraint.
Swan’s founders said no hiring — just the three of them. But they also set a goal: $30M ARR. That’s $10M per person. That constraint forced creativity. They didn’t ask, “Who do we hire next?” They asked, “How do we achieve this with what we’ve got?”
Takeaway: Don’t just “add AI.” Define a productivity metric or cost constraint that challenges the team to rewire how work gets done.
2. Don’t chase automation. Or AI agents. Chase orchestration.
Amos doesn’t see AI as a replacement for humans — he sees it as a way to scale himself. His “agentic swarm” supports every part of his GTM motion. He describes it as a tightly choreographed team—each agent playing a distinct role, passing the baton seamlessly to the next:
Shakespeare helps him turn raw thoughts into viral LI posts—generating 1M+ impressions a month.
The Observer listens for intent across 15K+ LI reactions and flags warm leads he’d never notice.
The Connector reviews 3K monthly connection requests and opens DMs with the right ones.
The Hunter analyzes 5K+ monthly website visitors who didn’t convert, qualifies them, and helps him engage the hottest ones personally.
The Gatekeeper enriches and filters 1.5K monthly inbounds and routes them to trial, demo, or waitlist.
The Prep Agent builds CRM-ready research briefs for 70 demos/week (stakeholder decision-making power, account research, and relevant case studies)
The Listener creates tasks, logs call summaries based on our sales framework, and drafts follow-up emails - so he can move straight to the next deal.
Takeaway: Start with orchestration. Pick one workflow — like demo prep or form routing — and explore how a human-plus-agent model could reduce time and increase output.
Think like a solo operator — even with a team.
Swan’s GTM team is one person: Amos. And he’s running the full-funnel. He’s replaced the question “what team do I need?” with “how do I make myself 100x more effective?” That’s a mindset shift any CEO can adopt.
Takeaway: Give one of your top ICs a “solo mission.” Challenge them to build a replicable agentic system for one GTM motion and scale it later.
This isn’t about starting from scratch. It’s about carving out a piece of your org — a team, a workflow, a motion — and proving what’s possible under constraint. Use the tools. But more importantly, shift the mindset.
Want more tactical examples of how to experiment with agent-led orchestration or challenge your team to solve with fewer resources?
Swan turned their entire Autonomous Business Playbook into a GPT that does exactly that.
No more PDFs. This one helps you apply the model: one agent, one system, one win at a time.
We’ve seen early wins in RevOps, support, and sales dev - all with zero headcount growth.
Want to try it?
Until next week!
I love this. What a great perspective - and actionable across all aspects of the business.