Hey!

Chris here. Welcome to Blueprint—the newsletter to help you build a winning engineering team.

I've been watching a pattern emerge across the teams I work with.

Everyone's using AI tools now, or at least starting to. Engineers are excited. Founders are excited. But when you look at how any given team is using them, every person has their own setup, rules, and configuration.

Nobody set out to build it that way. It just happened because nobody had a reason to think about it differently yet.

And for now, it mostly feels fine.

But here’s what I'm seeing on the other side of "fine." 👇

📒 DEEP DIVE

The Problem With Personal AI Workflows

Why AI-native teams need shared rules, shared context, and a single system that improves for everyone.

Every Man for Himself

Here's what the current state looks like in practice:

• Engineer A makes their own agents.md file

• Engineer B makes their own Claude.md

• Engineer C picks their own rules, adds their own plugins, sets up their own configuration

Nobody's doing anything wrong, they're just doing it differently.

What you end up with is 5 different setups living in 5 different people's heads. And the moment one of those people goes on vacation, that context goes with them. No good way for the team to operate as a unit.

The Moment It Breaks

Here's the scenario that makes this concrete.

Say I'm the one who's been deep in a problem. I've been working through why an agent is misbehaving, iterating on the fix, building up a full understanding of what's happening underneath. Then I go on vacation.

All of that context disappears with me. It lives in my head, my terminal, my session history. When I close the window, it's gone. My team is left starting from scratch.

In a shared workspace, that doesn't happen.

My team can see exactly what I typed. They can go in, take over the session, and read through everything in complete technical detail. If the agent broke in a specific way, they can see it, understand it, and make sure it doesn't happen again. Without me in the room.

We've never had that before.

Why Siloed Setups Can't Compound

There's a 2nd problem that's less obvious but just as damaging: When everyone is running their own setup, improvements stay isolated.

Say someone on your team figures out the right way to handle a specific problem. They tweak their rules, adjust their config, and it works. But that fix lives in their session. Nobody else benefits from it.

When the next person on the team hits the same issue, they solve it again from scratch.

There's no mechanism for the system to get collectively better.

When shared rules are in place, that changes.

One engineer figures out why the agent misbehaved and fixes the underlying rule, and that fix applies to everyone from that moment forward. You're not patching your own setup. You're improving the system for the whole team, permanently.

That's a fundamentally different kind of value.

What a Shared Workspace Looks Like

No product has fully solved this yet. I want to be upfront about that.

What I've been working toward is a shared workspace where a team of 2-5 people operates inside the same environment with the same rules, institutional knowledge baked into the system, and guardrails in place regardless of who's in the session.

The goal is simple: the system behaves consistently whether it's you in the session or someone else. And when something breaks, the knowledge of how it broke and how it got fixed is there for everyone.

What teams need is a shared operating environment with a thread running through all of it.

BEFORE YOU GO…

We're still in the early stages of what this looks like in practice. There are security questions to work through—not everyone on a team should have the same level of access, and admin-level visibility has to be controlled carefully.

But the direction is clear.

Teams that figure out how to operate inside a shared AI environment—with shared rules baked into the system—are going to move at a different speed than teams where everyone is still building their own setup in isolation.

Working with 1 or 2 sharp engineers inside a coherent system at full speed is one of the best feelings I've had in 23 years of building software.

The teams that don't think about this now are going to wake up with 5 different systems, no way to untangle them, and no idea why everything feels so inconsistent.

Talk soon,

Chris.