Skip to content

Instantly share code, notes, and snippets.

@jcohen66
Created February 4, 2026 20:31
Show Gist options
  • Select an option

  • Save jcohen66/92046b674a866acf74dcbe0526695e54 to your computer and use it in GitHub Desktop.

Select an option

Save jcohen66/92046b674a866acf74dcbe0526695e54 to your computer and use it in GitHub Desktop.
AI Declarative Spec #ai #pattern #declarative #spec #architecture #agent

If you are using AI model as the Oracle you are in the wrong frame of mind.

  1. Assign tasks don't ask questions. a) Describe the end state you want b) Provide the success criteria b) Let the system decide how to get there

  2. Accept imperfections and start to iterate a) Ralph Wiggum works because it produces broken code b) Make it retry until it fixes it c) Abandon's expectation that it should get it right the first time

  3. Invest in specification a) Invest in reviews b) Invest less in implementation c) Less time writing code d) More time defining what we want e) Evaluating whether we got there

When agents write the code, design becomes a bottleneck. Questions that slow you down: a) Architecture b) UX c) Composability d) What should this feel like? e) Do we have the right abstraction? f) Things Agents can't decide for you:
g) Require your Context, Vision, Taste

Move from coding to orchestration

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment