The Growth Squad of One

TL;DR (for skimmers)

• One person with AI now ships what used to take a squad.
• Raw experiment counts are down 60 percent,  but learning loops are faster because weak ideas die in simulation.
• Coordination‑only roles stall; vertical‑skill builders level‑up or get left behind.
• Tool‑hopping, data‑trust, and brand guardrails are still open problems—ripe for new products.

If you’ve got time for one coffee, read on.

1. The year everything compressed

For the past few months, we’ve been asking growth squads simple question: “What does your day-to-day look like now that you have LLMs?”

The answers sound a bit sci‑fi.

“I listened to a 30‑minute call, refined twice, and shipped a prototype and PRD in ten minutes.” — Hutton Jackson, Growth Director @ Tempo

To lightspeed and beyond, isn’t just changing the pace of growth teams, it’s an org‑chart earthquake. Lovable hit $80 M ARR with 35 people, blowing past milestones that once required triple‑digit headcount. Candu’s 2025 Product Growth Report found raw experiments were down 60 percent, not because growth teams are slowing down, but because most bad ideas never make it to production. LLM mock‑ups may be who killed them first; but formal growth teams themselves are also breaking down.

2. Meet the the AI-native growth PM

Think product manager, designer, engineer, and data analyst rolled into one:

Vibecoding? It’s live‑coding a feature while an AI pair‑programmer fills in boilerplate as you chat through the idea. The loop feels like Figma, GitHub Copilot, and Google Docs had a baby.

“Give me one prompt that creates everything, then let me tweak.” — Benjamin Franck, Sr. Product Manager, Boomi

3. More tests? No. Smarter bets

Growth culture used to worship A/B‑count. 2025 flips the metric: Decision time. If an LLM can help you develop five weak concepts in an afternoon, the one that reaches production can be much more fully realized—a pricing paywall or a full onboarding path, not a CTA change.

4. Frictions AI hasn’t solved

1. Tool‑hopping pain – briefs in ChatGPT, UI in Figma, launch in Candu, data in Amplitude. “After weeks of prompting, ChatGPT has accumulated so much product and brand context that switching would mean starting from scratch.  Context is a moat.” — Mike
2. Build a segment → Launch experiment → Analyze after – results still spans three dashboards
3. Trust – Benjamin has a healthy skepticism of AI suggestions‍
4. Skill gap – AI helps us play across more domains, but also challenges our skills. Growth PMs now craft prompts, tweak Flexboxes and write SQls in an afternoon.

5. Product Management in 2024 vs 2025

Important Caveat 

“A one‑human ‘experiment team’ doesn’t mean designers or engineers disappear. AI just kills the hand‑off treadmill—PMs can prototype fast, then bring in design and dev expertise to make a product that truly works.”  Kaelon Russell, Product Manager @ 15Five

6. Jobs, roles, and expectations

• Domain-light managers feel the squeeze. Coaching alone isn’t enough when one builder can do the loop.
• AI tilts leverage toward individual contributors. 
  
  • “AI is empowering PMs to work independently—as true ICs—while tools like Candu bundle design, product, engineering, and analytics into one surface.” — Benjamin Franck, Rivery  
• Coordination-only hires stall first. Open reqs freeze unless they add clear leverage.
• Role fluidity is the new normal. PMs sketch, code, analyze—AI fills the gaps they once waited on.
• Leaders raise the bar. If the tooling got 2× faster and you didn’t, something’s off.

“You build it, you own it… Velocity becomes your moat.” — Elena Verna

7. The 2025 toolkit in one slide

9. Open questions

1. When do you open the next head‑count? If one PM ships what a squad shipped last year, how do leaders and managers decide where to add?‍
2. How do you push AI up‑skilling across the org? Early adopters can’t carry the whole roadmap.
3. ‍Guardrails without bureaucracy? Code quality, data accuracy, brand safety—none can take a holiday.

Final take

The growth squad of one isn’t sci fi—it’s happening right now. The next edge won’t be another tool; it’ll be how fast we shed the habits that no longer serve us while keeping craft and quality intact.

Got a story about running a squad‑of‑one? Ping me—I’d love to swap notes.