git » homepage.git » commit 8db5022

diff --git a/md/AIAndRisk.md b/md/AIAndRisk.md
index ef820ba..e44118f 100644
--- a/md/AIAndRisk.md
+++ b/md/AIAndRisk.md
@@ -1,9 +1,9 @@
 # AIAndRisk
 
-A salaried engineer owns the knowledge in their head. That knowledge is the equity their employer pays for. Because the firm values their firsthand understanding of the artifacts, the engineer cannot safely outsource understanding to an AI. The minute they let Codex take risks on delivery or design, their personal equity erodes.
+A salaried engineer sells certainty. The firm buys their firsthand knowledge of the artifacts, and that human capital is the worker’s equity. In Austrian terms, the wage is payment for a service with low uncertainty. The moment the engineer lets Codex take risks on delivery or design, they erode the very knowledge the firm is paying for.
 
-Owners play a different game. An entrepreneur can let Codex explore risky options because the business, not the person, owns the asset. The owner is rewarded for throughput and speed, not for holding every detail. They can accept solutions they do not fully grasp as long as the system works and customers are served.
+Owners play the uncertainty game. Profit is the residual for bearing unknowable outcomes. An entrepreneur can push Codex into risky paths because the business, not the individual, owns the asset and the upside. They are rewarded for throughput and speed, not for holding every detail in their head.
 
-This makes AI upside asymmetric. The salaried contributor sees convenience and faster drafts, but the ceiling is low because trust in the artifacts must stay in their own head. The owner can push risk onto the model, harvest time savings, and keep moving without personally accumulating every detail.
+This makes AI upside asymmetric. For the salaried contributor, AI is a convenience with a low ceiling because trust in the artifacts must stay inside the worker. For the owner, AI is a capital good that extends production, absorbs uncertainty, and frees time without demanding personal comprehension of every step. The owner can accept imperfect knowledge if the system works and customers are served.
 
-If you are on salary, AI helps most when it expands your judgment: better questions, sharper reviews, clearer briefs. If you are an owner, AI helps when it absorbs execution and lets you take swings you would not have had time or staff to attempt. Both roles benefit, but only the owner can fully cash in on risk the AI is willing to shoulder.
+If you are on salary, use AI to widen judgment: better questions, sharper reviews, clearer briefs. If you own the risk, use AI to lengthen your production chain, accept experiments you would not fund with payroll, and harvest the profit from uncertainty that the model is willing to shoulder.