Prompt Versioning
Prompt versioning is treating prompts as code: every change is numbered, stored in a repository, tested against a fixed set of examples, and rolled back when a new version regresses.
No jargon, no math
In plain words, what is it?
A prompt looks like text, but it behaves like software: one edited word can change a hundred downstream answers. Versioning applies the discipline code has had for decades. Every prompt lives in a repository, not in a chat window. Each change creates a new numbered version with an author, a date and a diff. Before a version ships, it runs against a test suite: dozens or hundreds of fixed examples with known good answers, scored automatically. If the new version beats the old one on the tests, it goes live; if it regresses, it rolls back and the report says exactly which examples got worse. If a customer later disputes an answer, you can point to the exact version that produced it. Where it breaks: teams that keep prompts in production dashboards with no history. Nobody can say what changed, when, or why an output got worse last Tuesday.
Terms that live nearby.
Prompt
A prompt is everything you hand to a language model in one call: instructions, context and the question itself. The model answers by continuing exactly that text, and nothing else.
ReadPrompt Engineering
Prompt engineering is the practice of writing prompts so a language model performs a task reliably: clear instructions, the right context, worked examples of the desired output, and tests against real cases before you trust it.
ReadLLMOps
LLMOps is the discipline of running large language models in production: managing prompts, context, costs, latency and outputs with the same rigour MLOps applies to predictive models, adapted to models that answer in words and can say almost anything.
Read§ 03
Want this running in your stack?
We design and ship AI systems with permissions, logs and human approval. It starts with a 30-minute scoping call.