The Consortium for Information & Software Quality estimated poor software quality cost U.S. companies $2.41 trillion in 2022, with $1.52 trillion from operational failures alone

The Numbers Are Damning McKinsey estimates technical debt consumes 20–40% of engineering capacity at most large organizations

Velocity metrics displaced quality. Story points, burn-down charts, throughput — none measure whether software actually works. The Manifesto urged working software, but the focus shifted to speed

"Working software over documentation" hollowed out institutional memory. A 2023 GitLab survey reported only 12% of developers thought their codebases were well documented, making technical debt

"Responding to change" turned into a justification for never finishing features. Stanford researchers found scope creep institutionalized and relabeled as "sprint replanning," a leading

Those values aren't wrong on their own, but their practical interpretations became the problem

Individuals and interactions over processes and tools Working software over comprehensive documentation Customer collaboration over contract negotiation Responding to change over following a plan

Twenty-five years later Agile has grown into a $20B+ industry while the quality of delivered software is often declining The Four Principles

The Agile Manifesto was created in 2001 by 17 developers to rescue failing software projects. It helped for a while, but not forever.

Agile Has Broken Companies

The American people deserve confidence that every tool in government serves the public interest, and under President Trump no private company will ever dictate the terms of our national security

At the direction of the President, the Treasury is terminating all use of Anthropic products, including the use of its Claude platform, within the department

Software Factory has absorbed this complexity into its layer so software development doesn’t stop—the Factory stays resilient and continues operating Swapping out Anthropic is straightforward; shipping

That’s largely impractical right now, but as model quality converges it will become easier

No business should accept that level of risk—especially when AI is central to operations. Organizations need the ability to switch between models with minimal disruption.

If those diverge from the organization’s point of view, that creates major risk, as shown by the recent decision by a government department this weekend

Choosing a single model forces an organization to inherit that model maker’s institutional biases and quirks.

This is an important moment for all companies:

More downward pressure is probable. The spread between the equal-weight S&P 500 and the cap-weighted S&P 500 offers a simple gauge of the degree of repricing. Good luck to all the players!

That will likely drive deep cuts to stock-based compensation, since it materially impacts cash flow; the result would weaken employee retention and, in turn, growth, revenues and profits—creating a reflexive