The short answer

Start with RAG. It handles the overwhelming majority of business use cases, keeps your knowledge current, and lets answers cite their sources. Fine-tune only when you have measured a specific behavioural gap that retrieval cannot close.

This decision gets made backwards surprisingly often — teams fine-tune first because it sounds more sophisticated, then discover their model still cannot answer questions about last week's pricing.

What each actually does

RAG (retrieval)Fine-tuning
ChangesWhat the model can seeHow the model behaves
Good forFacts, documents, current dataFormat, tone, task patterns
Updating knowledgeRe-index — minutesRetrain — hours to days
CitationsNative — points at the sourceNot possible
Upfront costModerate engineeringData preparation plus training
Ongoing costRetrieval infra, larger promptsRetraining as behaviour drifts
Model portabilityWorks with any modelLocked to that model family

The row that decides most business cases is "updating knowledge". Your prices change, your policies change, your product catalogue changes. With retrieval that is a re-index. With fine-tuning it is a retraining cycle — and until you run it, your model is confidently telling customers last quarter's information.

Symptom-based diagnosis

Rather than choosing by architecture, choose by what is actually wrong:

SymptomLikely fix
Doesn't know our products or policiesRAG
Gives outdated informationRAG
Cannot cite sourcesRAG
Right facts, wrong tonePrompting first, then fine-tuning
Won't follow our output formatPrompting first, then fine-tuning
Doesn't understand our domain jargonRAG with a glossary; fine-tuning if persistent
Too verbose or too tersePrompting
Inconsistent across similar inputsFine-tuning, once prompting is exhausted

Notice how many rows resolve to prompting. A large share of "we need to fine-tune" conclusions are reached before anyone seriously iterated on the prompt — which costs nothing and takes an afternoon.

Try these in order

  1. Prompt engineering — free, immediate, and resolves more than expected. Exhaust it first.
  2. Few-shot examples — showing the model three good outputs often fixes format and tone entirely.
  3. RAG — for anything factual, current, or requiring citations.
  4. Prompt caching and context tuning — if cost or latency is the issue rather than quality.
  5. Fine-tuning — only for a measured, persistent behavioural gap the above cannot close.

The test before fine-tuning: can you write down the specific behaviour you want, and demonstrate with an evaluation set that prompting and few-shot examples do not achieve it? If not, you do not yet have evidence that fine-tuning is the answer — and it is the most expensive option to get wrong.

Where fine-tuning genuinely earns its cost

  • Strict output format at scale — when every response must conform exactly and prompting still produces variation.
  • A specialised task pattern — classification or extraction in a narrow domain where consistency matters more than flexibility.
  • Latency or cost pressure — a fine-tuned smaller model can outperform a larger prompted one on a narrow task, at lower cost per call.
  • Deeply domain-specific language — where terminology is genuinely unlike general usage and retrieval alone leaves the model confused.
  • Tone that resists prompting — a house style long prompts approximate but never quite hold.

The costs people underestimate

RAG costs

  • Chunking and indexing engineering
  • Vector store or search infrastructure
  • Larger prompts, so higher per-call token cost
  • Content hygiene — bad docs, bad answers
  • Retrieval quality tuning is real work

Fine-tuning costs

  • Training data curation — the largest hidden cost
  • Training compute
  • Retraining whenever behaviour must change
  • Lock-in to one model family
  • Evaluation to prove it improved anything

Training data curation deserves emphasis. Producing several hundred consistent, high-quality examples is significant expert time — and inconsistent examples actively make the model worse.

A practical recommendation

For nearly every business application — support agents, internal knowledge assistants, document processing, sales enablement — start with retrieval and disciplined prompting. Measure the result against a real evaluation set.

Revisit fine-tuning only when you can point at a specific, measured behavioural gap that survived prompting. At that point it is a targeted fix rather than an expensive guess.

Deciding between these for a real use case? Describe the symptom you are trying to fix — that usually settles it quickly. See our AI agent service, production RAG architecture, and why prompting is not enough.

Frequently asked questions

Yes, and for mature systems it is common — fine-tune for consistent format and tone, use retrieval for current facts. But start with retrieval alone. Adding fine-tuning before you have measured what retrieval cannot fix usually means paying for something that was never the problem.
Unreliably. Fine-tuning is effective at shaping behaviour, format and style. It is a poor mechanism for factual knowledge — the facts become baked in, cannot be updated without retraining, and the model still cannot cite where an answer came from.
Fewer examples than people expect for behaviour shaping — often hundreds rather than thousands — but they must be high quality and consistent. Poor training examples produce a model that is confidently wrong in a new way, which is worse than the base model.