Skip to content
dacard.ai
The point · The whitepaper

Twenty pagesSeven chaptersv1.4 · Feb 2026

Twenty pages on what changes when building is cheap and judgment is not.

Seven chapters, each listed with the claim it actually argues rather than a title you have to take on trust. Read the contents first and decide whether you disagree before you hand over an email address.

The whitepaper

The full case, in twenty pages. The email is how a conversation starts, not a gate.

No list, no sequence, no product to buy.

01The premiseOne shift, two refusals

One shift in where the cost sits, and two things this paper deliberately will not do.

Most writing on this subject argues that teams should adopt agents. This argues something narrower and more awkward: that adopting them without changing what you count makes a team faster at producing the wrong work.

What it argues

That the constraint moved. When an agent builds in an afternoon what took a team two weeks, the expensive act becomes choosing what to build and finding out whether it worked. Everything in the paper follows from that one shift.

What it will not do

It will not give you a maturity score, a rollout plan, or a tool to buy. It is a working argument written to be disagreed with, and the last chapter is the one most likely to be wrong.

02The contentsSeven chapters

Every chapter is listed with the claim it argues, so you can disagree with the argument before reading it.

Chapter 05 is where the paper stops describing and starts prescribing. Chapters 01 to 04 explain why it works, and 06 and 07 price what skipping it costs.

Contents · Twenty pages, seven chaptersv1.4 · Feb 2026
  1. 01p. 3

    The cost collapse

    Writing software stopped being the expensive part of building software.

  2. 02p. 6

    What got scarce

    Deciding what to build, and knowing whether it landed, did not get cheaper.

  3. 03p. 9

    The five-step loop

    People decide and judge, agents build, and step five has to finish before step one restarts.

  4. 04p. 13

    One name per call

    A recommendation is not a decision until a person puts their name on it.

  5. 05p. 16

    Landed, watch, stalled

    Effort is easy to count, so it becomes the report. Swap what you count.

  6. 06p. 19

    The rituals that survive

    Some meetings were asking a real question. Most were asking whether anyone was busy.

  7. 07p. 22

    Reading the ladder

    Five stages, described rather than scored, because a number invites debate about the number.