Lindsey Crowe

[Lindsey Crowe @linzcrowe]

This hits me directly. Vibe Kanban is part of my AI software factory workflow. I am also building Their Story and ShoSho, each touched by this shift differently. It changes how I will build new products going forward.

AI eats complexity for breakfast. The moat is thin.

[Lindsey Crowe @linzcrowe]

Vibe Kanban, a tool I use, just lost its commercial backing. Most users never paid and the team couldn't find a model that worked.

Companies like Atlassian charge per seat. But when AI does the work, agents sharing one key, who is the "user"? Seat-based pricing collapses.

[Lindsey Crowe @linzcrowe]

The dev team triangle is inverting.

PM at the top, engineers underneath used to be the shape. Now one developer with AI agents delivers faster than a PM can plan.

The bottleneck flipped. It used to be engineering capacity. Now it is the speed of deciding what to build.

[Lindsey Crowe @linzcrowe]

The dev team triangle is inverting.

PM at the top, engineers underneath used to be the shape. Now one developer with AI agents delivers faster than a PM can plan.

The bottleneck flipped. It used to be engineering capacity. Now it is the speed of deciding what to build.

[Lindsey Crowe @linzcrowe]

Even with solid test harnesses, Docker, and Storybook — AI-built features still do not look or feel right sometimes. Often not because the AI built it wrong, but because I did not specify things well enough.

So I am building AI-created walkthroughs.

[Lindsey Crowe @linzcrowe]

The bonus is that your team stops being human wikis and starts talking about strategy instead of "how do I query the logs again?"

Found a great way to share with human and AI alike? I'm keen to know!

[Lindsey Crowe @linzcrowe]

New rule when building with AI agents: ask the agent first, always.

Not your colleague or notes. The agent.

If your agent can't find the knowledge it needs, it won't stop and ask. It'll build something brilliant, working, tested — and wrong.

[Lindsey Crowe @linzcrowe]

Unsolved problem: agents spin up resources but don't always clean up. After a busy day — 100% CPU, memory maxed, everything grinding to a halt. Like telling the kids to tidy up and walking into 400 Lego pieces on the floor. 🙂

How are you handling multi-agent resource cleanup?

[Lindsey Crowe @linzcrowe]

How I ship with AI agents.
- Great skills and knowledge to upskill agents
- AI helps spec and design, human in the loop
- Automated tests, always

The last point is the real key. Strong quality gates, great tests, great results.

If you’re building like this, let’s compare notes.

[Lindsey Crowe @linzcrowe]

I run 4-8 AI coding agents in parallel daily. Biggest lesson so far: give each agent its own isolated environment (DB, services, containers). The difference between "my agent broke X again" and "my agent shipped a tested feature" is often just that.

[Lindsey Crowe @linzcrowe]

Keen to learn from others, help others learn, especially on the AI software factory angle.

[Lindsey Crowe @linzcrowe]

20 years in tech, and AI has me excited like never before. I am building a lean live-stream ecom platform (stealth), an AI software factory, and a memorial platform: strong tests, tight feedback loops, agents reviewing agents, humans in the loop. I will share as I go.

[Lindsey Crowe @linzcrowe]

Result:
Less context switching.
More flow.
Cleaner execution.
Not more automation — just clearer boundaries.
Curious: how do you separate “small” from “strategic” in your setup?

[Lindsey Crowe @linzcrowe]

Most dev teams lose flow on the small stuff.
A tweak.
A minor bug.
A “quick idea.”
It hijacks deep work.
My fix: split the system.

[Lindsey Crowe @linzcrowe]

My local coding agent + me handle the big work.
Architecture.
Refactors.
Anything with real consequences.
AI does the heavy lifting.
I guide direction and own responsibility.

[Lindsey Crowe @linzcrowe]

#tollipec @Toll_Group service is at an all time low. How can you mess up a simple package reschedule of the same package,3 times? And now because of that you'll likely miss the reschedule date? This is the core of your business, its what you do! Or did at least...

[Lindsey Crowe @linzcrowe]

@united ^MM's help via Twitter, 5 star! Help via your contact line, 2 star. Lost bag process, 0 start - needs drastic improvement

[Lindsey Crowe @linzcrowe]

@united very poor service - lost our bags with all our #wedding clothes, can't seem to properly track bags. Hard to have confidence in them

[Pew Environment @pewenvironment]

Junk in the ocean isn't just gross—it's dangerous: https://t.co/qVq5lY7ZL1