How has the cost and nature of software production changed with AI-generated code?
Answered at 8:00 in How to Live in Everyone Else's Future (with Shopify CEO Tobi Lütke) · ACQ2 by Acquired
How has the cost and nature of software production changed with AI-generated code?
Code used to be the most expensive thing to produce, but now with AI, a lot of code that glues together infrastructure can be treated as a first draft rather than a final product. This shift allows rapid iteration and experimentation—though truly critical systems still demand the same careful engineering and oversight they always have.
The economics of software development have fundamentally shifted. Where engineers once had to write production-ready code because iteration was costly, AI-generated code now enables a different approach: ship quickly, test thoroughly, and refine. As Tobi Lütke explains in this episode, this is "extremely healthy" for how companies approach building.
Infrastructure Code and the Speed-Quality Trade-off
Not all code carries the same risk profile. Infrastructure glue—the middleware, connectors, and scaffolding that ties systems together—benefits most from this new rapid-iteration model. Generated code can serve as a credible starting point, reducing the time teams spend on boilerplate and low-risk integration work, freeing human engineers for the deeper problems.
However, this permission to iterate fast does not extend universally. Code that underpins mission-critical functions—payment systems, security boundaries, core logic that directly affects users—still requires the same level of rigor, testing, and intentionality it always did. The shift is not about lowering standards everywhere; it's about redirecting engineering effort toward the systems where human judgment and accountability matter most.
"It is a privilege of a lifetime to be part of another platform shift."
Full answer and detailsRead the whole answer, key facts and every other question from this episode
In short
Code used to be the most expensive thing to produce, but now with AI, a lot of code that glues together infrastructure can be treated as a first draft rather than a final product.
Tobi Lütke · How to Live in Everyone Else's Future (with Shopify CEO Tobi Lütke) · ACQ2 by Acquired · answer at 8:00
Key takeaways
- This shift allows rapid iteration and experimentation—though truly critical systems still demand the same careful engineering and oversight they always have.
- The economics of software development have fundamentally shifted.
- Not all code carries the same risk profile.
- However, this permission to iterate fast does not extend universally.
Quote to remember
“It is a privilege of a lifetime to be part of another platform shift.”
— Tobi Lütke
The full answer · about 1 min
Code used to be the most expensive thing to produce, but now with AI, a lot of code that glues together infrastructure can be treated as a first draft rather than a final product. This shift allows rapid iteration and experimentation—though truly critical systems still demand the same careful engineering and oversight they always have.
The economics of software development have fundamentally shifted. Where engineers once had to write production-ready code because iteration was costly, AI-generated code now enables a different approach: ship quickly, test thoroughly, and refine. As Tobi Lütke explains in this episode, this is "extremely healthy" for how companies approach building.
Infrastructure Code and the Speed-Quality Trade-off
Not all code carries the same risk profile. Infrastructure glue—the middleware, connectors, and scaffolding that ties systems together—benefits most from this new rapid-iteration model. Generated code can serve as a credible starting point, reducing the time teams spend on boilerplate and low-risk integration work, freeing human engineers for the deeper problems.
However, this permission to iterate fast does not extend universally. Code that underpins mission-critical functions—payment systems, security boundaries, core logic that directly affects users—still requires the same level of rigor, testing, and intentionality it always did. The shift is not about lowering standards everywhere; it's about redirecting engineering effort toward the systems where human judgment and accountability matter most.
"It is a privilege of a lifetime to be part of another platform shift."
More answers from this episode
- 1:23What mental framework should a technology leader use to evaluate emerging AI capabilities?
- 4:40How should leaders design for technological futures they cannot fully predict?
- 9:00What role should platform abstractions play in software architecture?
- 16:00How can AI help merchants achieve their vision for their businesses?
- 20:00Why did the tech community largely overlook the moment AI systems passed the Turing test?
- 23:00What is the current state of consumer AR and VR technology adoption?