These four essays were published in The AI Journal between April and May 2026. They were written in sequence and they make one argument, which reads more clearly together than apart: the cost of doing the work is collapsing, and what survives the collapse is the ability to show that the work is right.
Copyright in all four remains with the author. They are linked here rather than reproduced.
AI Engine Optimisation: Why Your Finance Documents Are Failing AI, and What to Do About It
16 April 2026
Finance documents carry reasoning that a human reader reconstructs without noticing and a model cannot reconstruct at all. The fix is the one the web learned once already: make the implicit structure explicit, and write for the reader you now have.
The Execution Premium Collapsed. What Replaced It?
20 April 2026
Professional firms have charged for the ability to execute - to produce the model, the report, the analysis. When that ability stops being scarce, the premium has to move somewhere. This is an argument about where it goes.
If Execution Is Approaching Free, What Are We Charging For?
27 April 2026
The answer is trust, and trust is a build problem rather than a branding one. It is earned by architecture that lets a client check the answer for themselves: deterministic verification wrapped around a probabilistic core.
Same Thesis, Different Outcome: Why the Flywheel Only Compounds When the Work Is Re-engineered
11 May 2026
Seven venture firms converged independently on the same thesis about AI-native services. They will not get the same result from it. Automating an existing process leaves the process intact; the compounding advantage goes to whoever re-engineers the work itself.