4lbert Logo~/4lbert
HomeProfileSkillsPortfolioBlogGalleryContact
$ echo "© 2026 Torsten Albert · 4lbert.com"
ImpressumDatenschutz
Back to blog

Where Software Development Is Expensive

TorstenSeptember 17, 20264 min read

LLM coding agents make it cheap to turn ideas into working prototypes. That shifts the bottleneck from implementation to learning: instead of spending weeks perfecting plans, developers can challenge assumptions early, build small experiments, gather real feedback, and refine only what proves valuable. Extreme Programming’s focus on fast feedback, testing, and refactoring is more relevant than ever.

On this page
  • Extreme Programming fits this moment
  • “Garbage in, garbage out” still applies
  • Build prototypes to learn
  • Developers still own the outcome

For a long time, the expensive part of software development was implementation.

That shaped our process. We invested heavily in requirements, analysis, architecture, design documents, estimation, and planning before anyone started building. When a feature takes weeks or months to implement, that approach makes sense. Changing direction late is expensive.

LLM coding agents change the economics.

Given a clear enough task and a capable agent harness, an agent can inspect a repository, create a plan, implement a vertical slice, add tests, run checks, and document the result far faster than manual implementation.

That does not mean the agent replaces the developer.

It means the bottleneck is moving.

The difficult question is becoming less:

“Can we build this?”

And more:

“How quickly can we find out whether this is worth building?”

Extreme Programming fits this moment

Extreme Programming, or XP, focused on short feedback loops: small releases, automated tests, continuous integration, refactoring, pair programming, and close feedback from users.

Its main idea was not to predict every requirement in advance. It was to make change cheap once reality proved the original assumptions wrong.[1][2]

That idea is even more relevant when an agent can produce a first working version quickly.

Instead of following a long sequence of idea, analysis, design, implementation, and eventual feedback, we can move toward a tighter loop:

[ \text{hypothesis} \rightarrow \text{prototype} \rightarrow \text{feedback} \rightarrow \text{learning} \rightarrow \text{refinement} ]

The agent makes the prototype cheaper. XP gives us the discipline to learn from it without turning it into accidental production software.

“Garbage in, garbage out” still applies

A vague requirement can absolutely produce polished but useless code.

The agent may compile the application, generate tests, and create a convincing UI while still solving the wrong problem. “Garbage in, garbage out” remains true.

But I do not think the answer is always more up-front planning and increasingly detailed specifications.

The better approach is to let the agent challenge the idea before it implements it.

Before coding, ask it to identify assumptions, open questions, unclear decisions, risks, non-goals, and measurable success criteria. Then challenge the resulting plan instead of accepting it because it sounds plausible.

A tool such as Matt Pocock’s can help challenge a proposed plan and surface unanswered questions before implementation starts.[3][4]

PreviousWhen Mobile Banking Security Fails Without a Single BugNextAI Agents Need Boundaries

On this page

  • Extreme Programming fits this moment
  • “Garbage in, garbage out” still applies
  • Build prototypes to learn
  • Developers still own the outcome
/grill-me skill

The point is not to make the LLM responsible for product or architecture decisions. The point is to make hidden assumptions visible while changing direction is still cheap.

Build prototypes to learn

Not every important question can be answered in planning.

Will users trust the workflow? Does it save them time? Is the output useful with real-world data? Does the model behave safely when the context is incomplete?

Those are not design-document questions. They are experiment questions.

An agent makes it practical to build a narrow prototype early, evaluate it against real scenarios, and learn from the result. The prototype should be treated as evidence, not as a promise that the design is correct.

Tests, CI, reviews, permissions, and operational safeguards still matter. In fact, they matter more because fast generated code can create false confidence. A green test suite only proves something if the tests represent the behavior that users and the business actually need.

Developers still own the outcome

The agent can produce code quickly. It can propose alternatives, write tests, and point out missing cases.

But speed is not judgment.

Developers still need to decide whether the system solves a real problem, fits the architecture, respects data and security boundaries, and remains maintainable after the prototype phase.

That is the opportunity with LLM agents: not autonomous software development without engineers, but faster learning with engineers still in control.

The teams that benefit most will not be the ones generating the most code. They will be the ones that challenge assumptions early, build the smallest useful experiment, measure real outcomes, and refine only what proves valuable.

References [1] Extreme Programming (XP) at a Glance https://learn.microsoft.com/en-us/archive/blogs/jmeier/extreme-programming-xp-at-a-glance [2] What is Extreme Programming (XP)? - Agile Alliance https://agilealliance.org/glossary/xp/ [3] The /grill-me Skill - AI Hero https://www.aihero.dev/skills-grill-me [4] skills/skills/productivity/grill-me/SKILL.md at main ... - GitHub https://github.com/mattpocock/skills/blob/main/skills/productivity/grill-me/SKILL.md