This article is published in English.
Practical notes: The Part of RAG Where Quality Is Actually Won
Operable walkthrough of Practical notes: The Part of RAG Where Quality Is Actually Won: contracts, checks, and drop-in code slots for teams shipping this pattern.
Use this as an operator-facing rebuild of the ideas in “The Part of RAG Where Quality Is Actually Won”: clear stages, ordered code slots, and recovery notes that survive a handoff. The Overview stage works best when treated as a measurable surface. Capture one golden transcript, one failure case, and the rollback note before expanding scope. Prefer small, testable units over sprawling scripts. When a step fails, the failure should point at a single responsibility rather than a tangled pipeline.
Three kinds of chunks, three builders
For the Three kinds of chunks stage, define the inputs, the owner of the step, and the exit criteria before changing code. Operators should be able to re-run the step from a known checkpoint without guessing hidden state. Treat this stage as a contract between inputs and validated outputs. Name the artifacts, define success checks, and refuse silent partial completion. Cite the passages that actually grounded the answer. Without citations, operators cannot tell hallucination from an indexing gap.
The cleanup pass: deciding what the model never sees
For the The cleanup pass deciding stage, define the inputs, the owner of the step, and the exit criteria before changing code. Operators should be able to re-run the step from a known checkpoint without guessing hidden state. Record timings and token or query cost next to functional results. Cost visibility early prevents surprise bills when the path moves from demo to shared environments. Prefer structured outputs with schema validation over free-form prose when the next step is code or a tool call.
The label that changed everything
For the The label that changed stage, define the inputs, the owner of the step, and the exit criteria before changing code. Operators should be able to re-run the step from a known checkpoint without guessing hidden state. Keep configuration outside application code. Environment files, secret stores, and feature flags belong in one place operators can audit without reading the whole graph. Cite the passages that actually grounded the answer. Without citations, operators cannot tell hallucination from an indexing gap. For the The label that changed stage, define the inputs, the owner of the step, and the exit criteria before changing code. Operators should be able to re-run the step from a known checkpoint without guessing hidden state. Prefer small, testable units over sprawling scripts. When a step fails, the failure should point at a single responsibility rather than a tangled pipeline.
What the model actually sees
When working through the What the model actually stage, write down the contract first: required inputs, success signal, and what happens on partial failure. That checklist keeps later code changes honest. Treat this stage as a contract between inputs and validated outputs. Name the artifacts, define success checks, and refuse silent partial completion. Cache stable system instructions and tool schemas. Re-sending identical preamble is a common source of burn.
Then we embed
When working through the Then we embed stage, write down the contract first: required inputs, success signal, and what happens on partial failure. That checklist keeps later code changes honest. Record timings and token or query cost next to functional results. Cost visibility early prevents surprise bills when the path moves from demo to shared environments. Measure recall on a fixed question set before tuning prompts. Prompt churn rarely fixes a weak retrieval surface.
What a finished chunk actually is
When working through the What a finished chunk stage, write down the contract first: required inputs, success signal, and what happens on partial failure. That checklist keeps later code changes honest. Keep configuration outside application code. Environment files, secret stores, and feature flags belong in one place operators can audit without reading the whole graph. Measure recall on a fixed question set before tuning prompts. Prompt churn rarely fixes a weak retrieval surface. When working through the What a finished chunk stage, write down the contract first: required inputs, success signal, and what happens on partial failure. That checklist keeps later code changes honest. Prefer small, testable units over sprawling scripts. When a step fails, the failure should point at a single responsibility rather than a tangled pipeline.
text : "Enter amount and press F4."
(built + cleaned + section-prefixed)
retrieval_context : "Procedure step entering the amount
and pressing F4 during wire transfer
initiation." (the label)
chunk_type : TEXT
section_path : "Wire Transfer > Initiation >
Amount Entry" (Part 2)
preceding_heading : "Amount Entry"
table_headers : null (TABLE only)
page_start / end : 12 / 12
image_object_key : null (IMAGE only)
embedding_model : text-embedding-3-large
embedding_dim : 1536
embedding : [ ...1536 floats... ] (body+label)
text_search_vector: label weighted high, body lower
Paying for it on purpose
The Paying for it on stage works best when treated as a measurable surface. Capture one golden transcript, one failure case, and the rollback note before expanding scope. Treat this stage as a contract between inputs and validated outputs. Name the artifacts, define success checks, and refuse silent partial completion. Separate chunking policy from retrieval policy. Changing one should not force a rewrite of the other when quality metrics move.
Where we are, and what’s next
The Where we are and stage works best when treated as a measurable surface. Capture one golden transcript, one failure case, and the rollback note before expanding scope. Record timings and token or query cost next to functional results. Cost visibility early prevents surprise bills when the path moves from demo to shared environments. Separate chunking policy from retrieval policy. Changing one should not force a rewrite of the other when quality metrics move.
Operational checklist
When working through the Operational checklist stage, write down the contract first: required inputs, success signal, and what happens on partial failure. That checklist keeps later code changes honest.
Document the happy path and the recovery path together. Retries, human gates, and dead-letter handling are part of the product, not later polish.
Measure recall on a fixed question set before tuning prompts. Prompt churn rarely fixes a weak retrieval surface.
Pin dependency versions and record the image digest that ran the demo. Reproducibility beats tribal knowledge.
Prefer small, testable units over sprawling scripts. When a step fails, the failure should point at a single responsibility rather than a tangled pipeline.
Measure recall on a fixed question set before tuning prompts. Prompt churn rarely fixes a weak retrieval surface.
Before promoting the stack, freeze versions, capture a golden transcript for the critical path, and confirm rollback steps. Shared environments need rate limits, tenancy checks, and a clear owner for secret rotation. Prefer boring reliability over clever one-off demos.
Batch note for b5930b16a1d3: keep provider keys out of the repo, set a per-session token ceiling, and store transcripts next to the eval fixtures so later model swaps stay comparable.
When working through the hardening note 0 stage, write down the contract first: required inputs, success signal, and what happens on partial failure. That checklist keeps later code changes honest. Keep configuration outside application code. Environment files, secret stores, and feature flags belong in one place operators can audit without reading the whole graph.
Hardening detail 0/805: measure wall time, error class, and token spend for this note, then decide whether to keep the change based on a fixed question set rather than anecdote.
The hardening note 1 stage works best when treated as a measurable surface. Capture one golden transcript, one failure case, and the rollback note before expanding scope. Prefer small, testable units over sprawling scripts. When a step fails, the failure should point at a single responsibility rather than a tangled pipeline.
Hardening detail 1/805: measure wall time, error class, and token spend for this note, then decide whether to keep the change based on a fixed question set rather than anecdote.
For the hardening note 2 stage, define the inputs, the owner of the step, and the exit criteria before changing code. Operators should be able to re-run the step from a known checkpoint without guessing hidden state. Record timings and token or query cost next to functional results. Cost visibility early prevents surprise bills when the path moves from demo to shared environments.
Hardening detail 2/805: measure wall time, error class, and token spend for this note, then decide whether to keep the change based on a fixed question set rather than anecdote.
When working through the hardening note 3 stage, write down the contract first: required inputs, success signal, and what happens on partial failure. That checklist keeps later code changes honest. Document the happy path and the recovery path together. Retries, human gates, and dead-letter handling are part of the product, not later polish.
Hardening detail 3/805: measure wall time, error class, and token spend for this note, then decide whether to keep the change based on a fixed question set rather than anecdote.
The hardening note 4 stage works best when treated as a measurable surface. Capture one golden transcript, one failure case, and the rollback note before expanding scope. Treat this stage as a contract between inputs and validated outputs. Name the artifacts, define success checks, and refuse silent partial completion.
Hardening detail 4/805: measure wall time, error class, and token spend for this note, then decide whether to keep the change based on a fixed question set rather than anecdote.
For the hardening note 5 stage, define the inputs, the owner of the step, and the exit criteria before changing code. Operators should be able to re-run the step from a known checkpoint without guessing hidden state. Keep configuration outside application code. Environment files, secret stores, and feature flags belong in one place operators can audit without reading the whole graph.
Hardening detail 5/805: measure wall time, error class, and token spend for this note, then decide whether to keep the change based on a fixed question set rather than anecdote.
When working through the hardening note 6 stage, write down the contract first: required inputs, success signal, and what happens on partial failure. That checklist keeps later code changes honest. Prefer small, testable units over sprawling scripts. When a step fails, the failure should point at a single responsibility rather than a tangled pipeline.
Hardening detail 6/805: measure wall time, error class, and token spend for this note, then decide whether to keep the change based on a fixed question set rather than anecdote.
The hardening note 7 stage works best when treated as a measurable surface. Capture one golden transcript, one failure case, and the rollback note before expanding scope. Record timings and token or query cost next to functional results. Cost visibility early prevents surprise bills when the path moves from demo to shared environments.
Hardening detail 7/805: measure wall time, error class, and token spend for this note, then decide whether to keep the change based on a fixed question set rather than anecdote.
For the hardening note 8 stage, define the inputs, the owner of the step, and the exit criteria before changing code. Operators should be able to re-run the step from a known checkpoint without guessing hidden state. Document the happy path and the recovery path together. Retries, human gates, and dead-letter handling are part of the product, not later polish.
Hardening detail 8/805: measure wall time, error class, and token spend for this note, then decide whether to keep the change based on a fixed question set rather than anecdote.
When working through the hardening note 9 stage, write down the contract first: required inputs, success signal, and what happens on partial failure. That checklist keeps later code changes honest. Treat this stage as a contract between inputs and validated outputs. Name the artifacts, define success checks, and refuse silent partial completion.
Hardening detail 9/805: measure wall time, error class, and token spend for this note, then decide whether to keep the change based on a fixed question set rather than anecdote.
The hardening note 10 stage works best when treated as a measurable surface. Capture one golden transcript, one failure case, and the rollback note before expanding scope. Keep configuration outside application code. Environment files, secret stores, and feature flags belong in one place operators can audit without reading the whole graph.
Hardening detail 10/805: measure wall time, error class, and token spend for this note, then decide whether to keep the change based on a fixed question set rather than anecdote.
For the hardening note 11 stage, define the inputs, the owner of the step, and the exit criteria before changing code. Operators should be able to re-run the step from a known checkpoint without guessing hidden state. Prefer small, testable units over sprawling scripts. When a step fails, the failure should point at a single responsibility rather than a tangled pipeline.
Hardening detail 11/805: measure wall time, error class, and token spend for this note, then decide whether to keep the change based on a fixed question set rather than anecdote.
When working through the hardening note 12 stage, write down the contract first: required inputs, success signal, and what happens on partial failure. That checklist keeps later code changes honest. Record timings and token or query cost next to functional results. Cost visibility early prevents surprise bills when the path moves from demo to shared environments.
Hardening detail 12/805: measure wall time, error class, and token spend for this note, then decide whether to keep the change based on a fixed question set rather than anecdote.
The hardening note 13 stage works best when treated as a measurable surface. Capture one golden transcript, one failure case, and the rollback note before expanding scope. Document the happy path and the recovery path together. Retries, human gates, and dead-letter handling are part of the product, not later polish.
Hardening detail 13/805: measure wall time, error class, and token spend for this note, then decide whether to keep the change based on a fixed question set rather than anecdote.
For the hardening note 14 stage, define the inputs, the owner of the step, and the exit criteria before changing code. Operators should be able to re-run the step from a known checkpoint without guessing hidden state. Treat this stage as a contract between inputs and validated outputs. Name the artifacts, define success checks, and refuse silent partial completion.
Hardening detail 14/805: measure wall time, error class, and token spend for this note, then decide whether to keep the change based on a fixed question set rather than anecdote.
When working through the hardening note 15 stage, write down the contract first: required inputs, success signal, and what happens on partial failure. That checklist keeps later code changes honest. Keep configuration outside application code. Environment files, secret stores, and feature flags belong in one place operators can audit without reading the whole graph.
Hardening detail 15/805: measure wall time, error class, and token spend for this note, then decide whether to keep the change based on a fixed question set rather than anecdote.
The hardening note 16 stage works best when treated as a measurable surface. Capture one golden transcript, one failure case, and the rollback note before expanding scope. Prefer small, testable units over sprawling scripts. When a step fails, the failure should point at a single responsibility rather than a tangled pipeline.
Hardening detail 16/805: measure wall time, error class, and token spend for this note, then decide whether to keep the change based on a fixed question set rather than anecdote.