http://opakalex.github.io/tags/agents/Recent content in Agents on Alex Opak | Software ArchitectHugoen-usTue, 28 Jul 2026 00:00:00 +0000http://opakalex.github.io/posts/deterministic-agent-pipeline/Tue, 28 Jul 2026 00:00:00 +0000http://opakalex.github.io/posts/deterministic-agent-pipeline/LLMs are probabilistic by nature – probability buys flexibility, never 100% accuracy. So can a Talon + LLM agent pipeline promise same input, same output? The honest answer is a split: the decision core is provably deterministic, the LLM edge is only reproducible. Determinism comes from architecture, not from taming the model.