That 30% AI Boost Isn’t About Code. It’s About Communication.

I’ve built and scaled dev teams for over 15 years – here’s something I’ve learned the hard way: adding more developers doesn’t make them faster. Most of the time it makes them slower. 

Remember the handshake problem from maths class? If 5 people need to shake hands with everyone else it’s 10 handshakes. If 10 people do it that’s 45 handshakes.

Now think of the handshakes as communication channels. When you’ve got a 10 person team there are 45 potential communication channels. Meetings about meetings. Slack messages about who’s working on what. Merge conflicts. Code reviews that need three people to sign off.

Why 30% Efficiency Actually Matters

There are lots of numbers out there claiming that AI coding tools give you about 30% better efficiency per developer, focusing on the code output. That sounds nice. Here’s why it’s a much bigger impact than just the code:

Let’s say you’ve got a 10 person team. They need 45 communication channels to stay coordinated. You’re thinking about hiring 3 more people to handle the workload. If AI makes each developer 30% more productive you’ve got the same capacity as 13 people without hiring anyone. Your communication channels? Still 45. Not 78.

That’s the bit people miss –  you’re not just getting more code written. You’re avoiding the coordination tax that comes with growth.

Math

Let’s put numbers on it. You need the output of 13 developers. You’ve got two options:


Option 1: Hire 13 people
  • Communication channels: 78
  • Weekly coordination overhead: Maybe 20-25 hours of meetings, code reviews, alignment
  • Cost: 13 salaries plus recruitment

Option 2: 10 people with AI tools
  • Communication channels: 45
  • Weekly coordination overhead: Maybe 12 hours
  • Cost: 10 salaries plus AI tools (fraction of a salary)

Option 2 makes more sense. You’re not just 30% faster. You’ve avoided 33 extra communication channels. That’s 33 fewer ways for things to go wrong. The financial case is also compelling

What This Means in Practice

I’m not saying AI replaces developers. I’m saying it changes the economics of team size.

I’ve built teams to handle Shopify integrations, payment processors, inventory systems etc. Complex stuff. Before AI if I wanted to increase development velocity the only answer was to hire more developers. Now that isn’t the case anymore. 

Smaller teams communicate better. They ship faster. There’s less “I didn’t know you were working on that” and more “let’s build this.”

And when you do need to scale? Add 2 people instead of 5. Keep the team tight. Keep the communication clear.

The Bit Nobody Talks About

Measuring communication overhead is hard because it’s not just about the hours spent in meetings.

It’s the mental load of tracking who’s doing what. It’s the merge conflicts because someone didn’t know you’d touched that file. It’s the architectural decisions that need buy-in from 8 people instead of 4.

When AI handles the grunt work (boilerplate code, test generation, documentation), your developers can focus on the hard problems. The ones that actually need human judgement.

Fewer people need to be involved in those decision because the teams smaller. The teams smaller which means the communication channels are more manageable. 

Final Thought

If you’re running a team of developers and wondering why everything feels slow, look at your communication overhead first.

Then ask: could AI tools help us do more with fewer people? 
Not “let’s sack people.” 
But “could we achieve our goals without hiring the next 5 people we planned?”

I’ve been doing this for 15 years. I’ve built teams from 3 people to 30. The teams that ship best are the ones that stay small, stay focused, and don’t drown in coordination overhead.

That 30% efficiency gain isn’t just about code output. It’s about keeping your team small enough to actually work together.

Comments