The proportion nobody expects

Prompts are roughly 10–15% of a production AI system. The rest is retrieval, evaluation, integration, observability, cost control and failure handling — the engineering that determines whether the thing can be deployed at all.

The industry spent two years talking about prompting, which gave the impression it was the job. For anything reaching production, it is a small and shrinking part of it.

What a production system actually consists of

ComponentShare of effortWhat breaks without it
Data ingestion and preparation15–20%Garbage retrieval, garbage answers
Retrieval pipeline15–20%Right answer never reaches the model
Prompting10–15%Inconsistent format and tone
Integration with your systems15–20%Cannot access real data
Evaluation10–15%No basis for a go-live decision
Guardrails and safety10%Cannot pass security review
Observability and cost control8–12%Silent degradation; runaway bills

Data preparation being the largest item surprises people. Extracting text cleanly from real documents, preserving structure, deduplicating, and keeping content current is unglamorous work that caps everything downstream. No prompt recovers from a corpus where half the PDFs were extracted badly.

What prompting genuinely fixes

  • Output format and structure.
  • Tone and register.
  • Instruction adherence — what to do and what to avoid.
  • Reasoning approach for a given task.
  • How to behave when information is insufficient.

What prompting cannot fix

  • Missing information. If retrieval did not find it, no instruction conjures it.
  • Bad source content. Outdated documentation produces outdated answers, confidently.
  • Absent system access. The model cannot look up an order without a tool that does.
  • Unmeasured quality. Prompting without evaluation is changing things and hoping.
  • Cost problems. A prompt cannot reduce a bill driven by oversized retrieved context.
  • Security exposure. "Ignore malicious instructions" is not a security control.

Teams stuck iterating endlessly on prompts are almost always fighting a retrieval or data problem. The symptom presents at the prompt; the cause is upstream.

Diagnosing where the real problem is

SymptomUsually blamed onUsually caused by
Wrong or vague answersThe promptRetrieval missing the content
Outdated informationThe modelStale source documents
Inconsistent qualityThe modelNo evaluation, so drift is invisible
High costThe model's pricingRetrieving far more context than needed
Slow responsesThe modelSequential calls; oversized prompts
Wrong tool calledThe modelAmbiguous tool descriptions

The diagnostic that resolves most arguments: take a failing case and manually paste the correct information into the prompt. If the model then answers well, your problem is retrieval — and no amount of prompt tuning will fix it. If it still fails, the prompt or the model is genuinely the issue.

What to build alongside the prompt

  1. An evaluation set — real cases with expected outcomes. Build it first; it makes every later decision measurable.
  2. Retrieval scored separately from generation, so failures are attributable.
  3. Tracing — what was retrieved, what was sent, what came back, what it cost.
  4. Refusal and handoff behaviour, tested deliberately.
  5. Cost attribution per feature.
  6. Regression gating — no prompt change ships without the evaluation passing.

What to look for when hiring

  • Software engineers who understand LLM behaviour — not prompt specialists without engineering depth.
  • Experience with retrieval systems and search relevance.
  • Comfort with evaluation and measurement as a discipline.
  • Systems integration capability, since most of the work is connecting things.
  • Observability instincts — knowing what to instrument before it breaks.

Iterating on prompts without the results improving? Describe the failure pattern — the cause is usually upstream. See our AI agent service, RAG architecture, and evaluation frameworks.

Frequently asked questions

Not going away, but shrinking in relative importance. Models follow instructions better each generation, which reduces the effort spent coaxing behaviour. The surrounding engineering — retrieval, evaluation, observability, cost control — does not shrink at all.
Hire software engineers who understand LLM behaviour, not prompt specialists. The work is overwhelmingly systems engineering. A prompt specialist with no engineering background will not be able to build the 90% that surrounds the prompt.
On a typical production build, somewhere under 15% of total effort — with most of that spent early, then revisited when evaluation reveals a gap. If prompts are consuming most of your project, something upstream is broken.