AI Is Building Itself. Your Execution Moat Is Dying
When execution goes near-zero, your real business is what AI cannot copy.
Claude now writes more than 80% of the code merged inside Anthropic. Their engineers are shipping roughly 8x more code per quarter than before.
If your business sells execution, your price floor just moved. Watch the full breakdown for the five founder moves — this issue gives you the audit underneath them.
The video is the full story. Below is the layer underneath — plus one thing only newsletter subscribers get.
What They Actually Said
More than 80% of the code merged into their codebase is written by Claude.
It’s been 5 months since I last wrote any code myself.
On days where everything works well, I can’t help but think nothing I do matters.
The Breakdown
The scary part is not that AI can write code. We already knew that.
The scary part is that the companies building frontier AI are now using AI to accelerate the building of frontier AI. That changes the speed limit. The machine is no longer just doing tasks inside the economy. It is improving the factories that make the next machines.
For founders, this means the old moat of “we can deliver high-quality work reliably” is weakening. Reliability used to be expensive. Speed used to require headcount. Now both are being compressed by agents. The value is moving away from execution and towards direction, taste, trust, data, and distribution.
The Insight
You might think the threat is AI replacing your team.
Actually — the opposite. The bigger threat is a smaller competitor who keeps their team, puts agents underneath every role, and suddenly delivers with the speed of a company 10x their size. They do not need to be smarter than you. They only need to remove the human routing layer before you do.
The Action
This week, take your last 10 client projects or internal deliverables and mark every step as either “execution” or “judgment”. Execution is production, formatting, research, admin, first drafts, QA passes, reporting, scheduling. Judgment is taste, client context, strategic calls, risk, positioning, final approval. Anything marked execution more than 3 times becomes your first agent workflow to build or delegate to AI before your next hire.
The Execution Line Audit
Only in the newsletter — not in the video.
Use this to find the exposed parts of your company before the market does:
- List your 5 most repeated workflows from the last 30 days.
- For each one, write the exact output produced.
- Mark every step as E for execution or J for judgment.
- Circle any workflow with more than 60% E steps.
- Ask: “Could a trained agent produce the first version of this?”
- If yes, create one reusable prompt, checklist, or SOP this week.
- Keep humans on the final judgment layer only.
- Store the improved process in one shared folder, not someone’s head.
- Repeat weekly until every role has an agent underneath it.
Get this every week.
Every week we translate the top AI news into clear strategy — and one practical thing to actually do this week. For founders, builders, and operators. Plus a subscriber-only template, checklist, or teardown in every issue.