It’s sort of difficult to understand why this is even a question - LLM-based / judgment dependent workflows vs script-based / deterministic workflows.<p>In mapping out the problems that need to be solved with internal workflows, it’s wise to clarify where probabilistic judgments are helpful / required vs. not upfront. If the process is fixed and requires determinism why not just write scripts (code-gen’ed, of course).
> We still start all workflows using the LLM, which works for many cases. When we do rewrite, Claude Code can almost always rewrite the prompt into the code workflow in one-shot.<p>Why always start with an LLM to solve problems? Using an LLM adds a judgment call, and (at least for now) those judgment calls are not reliable. For something like the motivating example in this article of "is this PR approved" it seems straightforward to get the deterministic right answer using the github API without muddying the waters with an LLM.
There is a third option, letting AI write workflow code:<p><a href="https://youtu.be/zzkSC26fPPE" rel="nofollow">https://youtu.be/zzkSC26fPPE</a><p>You get the benefit of AI CodeGen along with the determinism of conventional logic.