GraphRAG in Production — Part 1: Knowledge Graph Build Templates — 42 Templates & Checklists for Question Classification, Knowledge Graph Fundamentals, Extraction Pipelines, Ontology Design, Graph Store Selection, and Graph Indexing

## Five relevant passages. No answer. Forty minutes of manual link-following.

At 02:17 on a rainy Tuesday, the operations lead at a freight company asked the question that opens the source book: *which supplier outage delayed which shipments, and which customers are still waiting because of it?* The vector assistant answered instantly and confidently — five beautifully relevant passages, none of which, together or alone, answered the question, because the answer lived in no single passage. It lived in a path: supplier to factory, factory to shipment, shipment to customer, checked at every hop. **GraphRAG in Production — Part 1: Knowledge Graph Build Templates** is the working build for the first half of that discipline: 42 copy-paste templates and checklists distilled from Chapters 1–6 of the book — decide whether your problem is a graph problem at all, lay the fundamentals that survive real data, extract facts under governance, design the schema as the product it is, choose the ground the graph stands on, and index it so a question can actually enter.

Every template carries the book's tested positions built in: questions classified lookup / relational / global and priced by error cost before any architecture is chosen; canonical IDs that are opaque, durable, and owned, with names demoted to properties; provenance required on every fact — a load without a source document fails loudly; extraction as seven governed stages where the model proposes and the pipeline disposes, with span-validated JSON contracts, dual-threshold entity resolution, and a human review queue; an ontology designed backwards from 30–60 competency questions, layered core / domain / application, versioned additively, and enforced by validation gates with specific rejection reasons; store selection that starts from five defining traversals, tests supernodes on production-shaped data, and scores finalists with transparent weights; and an index of node/edge/text-unit embeddings, hierarchical Leiden communities, and community summaries gated on sampled faithfulness. Where the book gives a number — its chunk range of a few hundred to about a thousand tokens, its illustrative cost model of $100–$300 per thousand documents, its fictional worked cases — the template carries the book's number and labels it as the book's, illustrative where the book labels it illustrative. Bracketed fields are yours to fill; nothing here invents a threshold for your corpus.

### What's inside — 42 templates & checklists in 7 sections

**1. Decide First — Is Your Problem a Graph Problem? (6)** — Question-Class Audit · Vector-vs-Graph Behaviour Comparison Sheet · Path-Answer Trace Worksheet · Two-Plane Architecture & Trust-Boundary Sheet · Indexing Cost Estimate Worksheet · Graph-or-Not Decision & 30-Day Pilot Plan

**2. Graph Fundamentals — Identity, Direction, Provenance, Time (6)** — Triple & Property Design Sheet · Property Graph vs RDF Selection Sheet · Naming & Vocabulary Control Sheet · Canonical ID & Merge-Log Sheet · Provenance & Temporal Contract Sheet · Granularity, Schema-Regime & Health Sheet

**3. Extraction — From Documents to Governed Graph Facts (6)** — Seven-Stage Pipeline Map & Stage-Metrics Sheet · Extraction Chunking Specification · Extraction Prompt & JSON Contract Sheet · Gleaning & Precision–Recall Decision Sheet · Entity Resolution & Human Review Queue Sheet · Extraction Cost Model, Failure Table & Incremental Operation Sheet

**4. The Schema Is the Product — Ontology Design (6)** — Competency-Question Design Worksheet · Three-Layer Schema Map · Naming, Direction & Signature Register · Versioning, Deprecation & Change-Proposal Sheet · Ontology Validation-Gate Specification · Over/Under-Modeling Diagnosis & Two-Week Design Plan

**5. Choosing the Ground — Graph Store Selection (6)** — Five Defining Traversals Worksheet · Native vs Layered Assessment & Supernode Test Sheet · Store Shortlist Comparison Sheet · Per-Store Modelling Sheet · Weighted Scoring Worksheet · Operations, Exit & No-Store Decision Sheet

**6. Indexing the Graph — Embeddings, Communities, Summaries (6)** — Index-Layer Sizing Sheet · Node, Edge & Text-Unit Embedding Specification · Community Detection & Resolution Tuning Sheet · Community Summary Template & Quality Gate · Indexing Cost Curve & Re-Indexing Plan · Index Failure & Health-Monitor Sheet

**7. Part 1 Gates & Working Artefacts (6)** — Decision Gate (Chapter 1 checklist) · Foundation Gate (Chapter 2 checklist) · Extraction Gate (Chapter 3 checklist) · Schema Gate (Chapter 4 checklist) · Store Selection Gate (Chapter 5 checklist) · Part 1 Build Gate — graph indexing checklist & full sign-off (the book's six chapter checklists consolidated across Items 37–42, 48 gate lines)

### Who this is for

AI and platform engineers building retrieval over connected data, search and knowledge engineers moving past a vector-only baseline, and technical leads who must decide — with evidence — whether a knowledge graph earns its indexing bill. The source book assumes you build production systems; it teaches the graph build, not programming. If your highest-value users ask relational questions your assistant answers by stitching true passages into false chains, start at Template 1.

### What this pack is not

- **Not Part 2.** Hybrid retrieval, query planning, evaluation, and scale are the book's Chapters 7–12 and Appendices A–B. They live in the Part 2 companion pack; nothing here restates them, and this pack is complete and useful without them.

- **No runnable code, pipeline, or store setup.** The book's Python sketches (a triple store, a property graph, an extraction core, an ontology validator, community detection) are simplified, structurally honest teaching shapes. This pack ships the templates that specify, govern, and operate the build — not a working extraction pipeline, graph database installation, or indexing service.

- **No vendor selection for you.** Neo4j, Amazon Neptune, ArangoDB, TigerGraph, and JanusGraph are compared as the book compares them, and the scoring worksheet runs on YOUR workload evidence. No store is installed, configured, licensed, or endorsed here.

- **No benchmark or results claims.** The book's worked figures — traversal times, graph sizes, acceptance rates in its fictional cases — are the book's illustrative examples, labelled as such in every template that carries them. This pack promises no accuracy, latency, or cost outcome for your system.

- **No guarantee a graph is right for you.** The book's own position, built into Templates 1 and 6: FAQ-style corpora, relation-free text, and clean relational data need no graph, and an indexing plane that cannot be funded is a reason to stop. These templates include the honest exits, not just the build.

- **Not a vector-RAG pack.** Chunking for retrieval, embedding-model selection, and vector-store operations are other books' shelves (the book itself points to them). This pack covers chunking only as extraction chunking — sized for relational completeness — and embeddings only as graph entry points and text units.

### Companion products

**GraphRAG in Production — Part 2** is this pack's sibling, built from Chapters 7–12 and Appendices A–B of the same book (hybrid retrieval, query planning, evaluation, scale): a companion for when the built graph must be queried and scaled — never a requirement for this pack to work. This pack stands alone.

*Note: the $99 Architecture Templates Bundle currently contains Products #1–#3 only. This pack launches standalone; no bundle inclusion is promised for this product.*

Built by the author of **GraphRAG in Production**, from the build discipline behind production knowledge-graph systems.

### About the author

Naveen Kumar Josyula is the author of 35 AI books on Amazon KDP and the CEO and founder of AISoltrics Technologies Inc. He is the author of the source book, *GraphRAG in Production*, from which every template in this pack is distilled. A separate KDP book line also carries a GraphRAG title; this pack is distilled from the manuscript named above.

### Support promise

If anything in this file is wrong or unclear, email support@aisoltrics.com before you judge it — we fix verified issues within one business day and send you the corrected file free. (You can also reach the founder's office at aadhya@aisoltrics.com.) Updates to this product are free to past buyers via your library. Sold by AISoltrics Technologies Inc.