← Back to articlesHow models work

Prompting, RAG or fine-tuning: where should you start?

First identify whether the gap is instructions, current information or a consistent specialized behavior.

The short answer

Prompts clarify tasks, retrieval supplies updatable material, and fine-tuning involves training and evaluation. Uploading a document is not fine-tuning.

Try it in three steps

  • Describe one measurable failure.
  • Retry with clearer instructions and appropriate information.
  • Only then consider training data, cost and evaluation.

A practice prompt

Before you start

Classify the problem first without buying or training a model.

Practice material (fictional)

A: Column headings vary. B: The updated manual is ignored. C: Sufficient approved examples exist but a specialized format remains unreliable.

Replace placeholders in the prompt with this material, or append the material if there are no placeholders. Reading time excludes doing the exercise.

Our manual changes often and answers need sources. Compare what prompts, RAG and fine-tuning address without declaring an untested winner.

What to look for (not a recorded test result)

Clarify A's format; inspect B's sources and retrieval; evaluate training and tests for C without assuming fine-tuning is necessary.

If it goes wrong

If one policy changed, update the retrievable material before considering retraining.

Check your result

Your choice addresses a specific failure and can be tested again on the same questions using the same steps.

References and scope

The links below support the concepts or product navigation. The steps and exercises are our teaching suggestions, not official certification or measured customer outcomes. Product interfaces and availability may change. References checked on 2026-10-17.

Start with one task

Describe the current process and its slowest step. The initial conversation is free.

Let’s talk ↗
Need help? See small-scale implementation services ↗

Keep reading