---
title: How 300+ AI Builders Combine Knowledge Graphs and LLMs
description: Hundreds of teams built enterprise AI apps with no set architecture. The strongest landed on the same design principles. Here's what they are.
image: https://info.tigergraph.com/hubfs/The%20Connected%20Intelligence%20Experiment.jpg
---

![logoblack](https://info.tigergraph.com/hubfs/logoblack.svg)

- [Product](https://www.tigergraph.com/product/)
- [Solutions](https://www.tigergraph.com/solutions/)
- [Developers & Data Scientists](https://www.tigergraph.com/developers/)
- [Resources](https://www.tigergraph.com/resources/)
- [Services & Support](https://www.tigergraph.com/support/)
- [Company](https://www.tigergraph.com/company/)

[Contact Us](https://www.tigergraph.com/contact/)

- [Product](https://www.tigergraph.com/product/)
- [Solutions](https://www.tigergraph.com/solutions/)
- [Developers & Data Scientists](https://www.tigergraph.com/developers/)
- [Resources](https://www.tigergraph.com/resources/)
- [Services & Support](https://www.tigergraph.com/support/)
- [Company](https://www.tigergraph.com/company/)

[Contact Us](https://www.tigergraph.com/contact/)

![](https://info.tigergraph.com/hubfs/Landing%20Page_The%20Connected%20Intelligence%20Experiment.jpg)

Report

# **The Connected Intelligence** Experiment

What 300+ AI builders learned about combining knowledge graphs and LLMs

Give hundreds of teams the same tools, a set of enterprise problems and no prescribed architecture. What do they build? In this experiment, more than 300 builders did exactly that on TigerGraph. Working independently, the strongest teams kept arriving at the same design choices for connecting knowledge graphs to large language models. This report lays out those principles and the reasoning behind them.

### Download the free report

![Clients\_Logo-Nov-21-2025-08-20-07-5352-PM](https://info.tigergraph.com/hubfs/Clients_Logo-Nov-21-2025-08-20-07-5352-PM.svg)

![Clients\_Logo-14](https://info.tigergraph.com/hubfs/Clients_Logo-14.svg)

![Clients\_Logo-13](https://info.tigergraph.com/hubfs/Clients_Logo-13.svg)

![Clients\_Logo-12](https://info.tigergraph.com/hubfs/Clients_Logo-12.svg)

![Clients\_Logo-11](https://info.tigergraph.com/hubfs/Clients_Logo-11.svg)

![Clients\_Logo-10](https://info.tigergraph.com/hubfs/Clients_Logo-10.svg)

![Clients\_Logo-9](https://info.tigergraph.com/hubfs/Clients_Logo-9.svg)

![Clients\_Logo-8](https://info.tigergraph.com/hubfs/Clients_Logo-8.svg)

![Clients\_Logo-7](https://info.tigergraph.com/hubfs/Clients_Logo-7.svg)

![Clients\_Logo-6](https://info.tigergraph.com/hubfs/Clients_Logo-6.svg)

![Clients\_Logo-5](https://info.tigergraph.com/hubfs/Clients_Logo-5.svg)

![Clients\_Logo-4](https://info.tigergraph.com/hubfs/Clients_Logo-4.svg)

![Clients\_Logo-3](https://info.tigergraph.com/hubfs/Clients_Logo-3.svg)

![Clients\_Logo-2](https://info.tigergraph.com/hubfs/Clients_Logo-2.svg)

![Clients\_Logo-1](https://info.tigergraph.com/hubfs/Clients_Logo-1.svg)

## Design principles the **strongest teams shared**

![8429366\_ecommerce\_shopping\_online\_attention\_magnifier\_icon 1](https://info.tigergraph.com/hubfs/8429366_ecommerce_shopping_online_attention_magnifier_icon%201.svg)

### Model the relationships, not just the documents

Top builds defined the entities and links in their domain before writing prompts.

![8388125\_2business\_finance\_office\_organize\_structure\_icon 1](https://info.tigergraph.com/hubfs/8388125_2business_finance_office_organize_structure_icon%201.svg)

### Let the graph decide what the model sees

Graph queries selected a small, connected set of facts, keeping prompts short and focused.

![8463505\_human\_anatomy\_body\_brain\_organ\_icon 1](https://info.tigergraph.com/hubfs/8463505_human_anatomy_body_brain_organ_icon%201.svg)

### Use vectors and graph together

Semantic search found the starting points and graph traversal filled in the context around them.

![8441437\_essential ui\_app\_web\_protection\_shield\_icon 1](https://info.tigergraph.com/hubfs/8441437_essential%20ui_app_web_protection_shield_icon%201.svg)

### Keep the reasoning visible

Teams that exposed the path behind each answer found errors faster and earned trust sooner.

## What the report **covers**

### How the experiment was set up and how builds were scored

### The design principles that kept appearing in the strongest builds

### Common mistakes that held weaker builds back

### Reference patterns for connecting a knowledge graph to an LLM

### How to apply the principles to your own AI roadmap

## Who should **read it**

Architects and AI leads designing LLM applications on enterprise data who want patterns proven by many independent teams rather than a single vendor reference design.

## Knowledge graphs and LLMs, **answered**

### How do knowledge graphs improve LLM accuracy?

A knowledge graph gives an LLM structured facts and the relationships between them. Instead of guessing from loosely related text, the model works from a connected view of entities, which improves accuracy on complex questions, reduces hallucination and makes it possible to trace how an answer was formed.

### How do you use a knowledge graph with an LLM?

Queries run against the knowledge graph, often alongside vector search, to assemble the entities and relationships relevant to a question. That connected context is passed to the LLM in the prompt, and the model generates an answer grounded in those specific facts rather than in its training data alone.

### Why run an open experiment instead of a benchmark?

A fixed benchmark tests one architecture. Leaving the design open shows which choices teams make on their own when solving real problems. When hundreds of independent builds converge on the same patterns, that is strong evidence those patterns work.

![](https://info.tigergraph.com/hubfs/5_CTA%20section.jpg)

## Build on what **already works**

Get the full set of design principles from 300+ builds. To try them on your data, book a TigerGraph demo or start free in Savanna.

[Download the free report](https://info.tigergraph.com/the-connected-intelligence-experiment?hsLang=en-us#download-form) [Contact us](https://www.tigergraph.com/contact)

### → Download the free report

## Related **reports**

[**GraphRAG in Practice** GraphRAG use cases and results from 300+ enterprise AI builds →](https://info.tigergraph.com/graphrag-in-practice?hsLang=en-us) [**GraphRAG Builder Playbook** How to build GraphRAG that holds up in production, based on 300+ enterprise AI builds →](https://info.tigergraph.com/graphrag-builder-playbook?hsLang=en-us)

## Products

### [TigerGraph Savanna](https://www.tigergraph.com/savanna/)

- [FAQ](https://www.tigergraph.com/savanna-faq/)
- [Pricing](https://www.tigergraph.com/savanna-pricing/)

### Language and Tools

- [Solution Kits](https://www.tigergraph.com/solution-kits/)
- [GSQL Query Language](https://www.tigergraph.com/gsql/)
- [OpenCypher Query Language](https://www.tigergraph.com/opencypher-support/)
- [Graph Studio](https://www.tigergraph.com/graphstudio/)
- [Insights](https://www.tigergraph.com/insights/)
- [Connectors](https://www.tigergraph.com/connectors/)

### [TigerGraph DB](https://www.tigergraph.com/tigergraph-db)

- [Compare Editions](https://www.tigergraph.com/comparison-of-tigergraph-editions/)

### [Graph Data Science](https://www.tigergraph.com/graph-data-science-library/)

- [Graph Data Science Library](https://www.tigergraph.com/graph-data-science-library/)

## Solutions

### [Solutions](https://www.tigergraph.com/solutions/)

- [Overview](https://www.tigergraph.com/solutions/)

### Improve Operations

- [Supply Chain Analysis](https://www.tigergraph.com/solutions/supply-chain-analysis/)
- [Network Optimization](https://www.tigergraph.com/solutions/network-it-and-cloud-resource-optimization/)
- [Energy Management System](https://www.tigergraph.com/solutions/energy-management-system/)

### [Increase Revenue](https://www.tigergraph.com/solutions/product-service-marketing/)

- [Product and Service Marketing](https://www.tigergraph.com/solutions/product-service-marketing/)
- [Customer 360](https://www.tigergraph.com/solutions/real-time-customer-360mdm/)
- [Entity Resolution](https://www.tigergraph.com/solutions/entity-resolution/)
- [Recommendation Engine](https://www.tigergraph.com/solutions/recommendation-engine/)

### Manage Risk

- [Fraud Detection and Protection](https://www.tigergraph.com/solutions/fraud-detection/)
- [Cybersecurity Threat Detection](https://www.tigergraph.com/solutions/cybersecurity-threat-detection/)
- [Anti-Money Laundering (AML)](https://www.tigergraph.com/solutions/anti-money-laundering-aml/)
- [Risk Assessment and Monitoring](https://www.tigergraph.com/solutions/risk-assessment-and-monitoring/)

### By Industry

- [Advertising, Media, Entertainment](https://www.tigergraph.com/solutions/advertising-media-and-entertainment/)
- [Financial Services](https://www.tigergraph.com/solutions/financial-services/)
- [Healthcare & Life Sciences](https://www.tigergraph.com/solutions/healthcare-and-life-sciences/)

### Foundational

- [AI & Machine Learning](https://www.tigergraph.com/solutions/ai-and-machine-learning/)
- [Time Series Analysis](https://www.tigergraph.com/solutions/time-series-analysis/)
- [Geospatial Analysis](https://www.tigergraph.com/solutions/geospatial-analysis/)

## Developers and Data Scientists

### [Download](https://dl.tigergraph.com/)

- [TigerGraph DB](https://dl.tigergraph.com/)
- [TigerGraph Docker Image](https://dl.tigergraph.com/)
- [GSQL Client](https://dl.tigergraph.com/)

### Developers

- [Documentation](https://docs.tigergraph.com/home/)
- [Developer Hub](https://www.tigergraph.com/developers/)
- [Ecosystem](https://www.tigergraph.com/ecosystem/)
- [Community Forum](https://dev.tigergraph.com/forum/)

## Services and Support

### Services and Support

- [TigerGraph University](https://www.tigergraph.com/training/)
- [Training & Certifications](https://www.tigergraph.com/tigergraph-university-program/)
- [Support](https://www.tigergraph.com/support)
- [Production Guidelines](https://www.tigergraph.com/production-guidelines/)

## Resources

### [Blog](https://www.tigergraph.com/blog/)

- [TigerGraph Blog](https://www.tigergraph.com/blog/)

### [Resources](https://www.tigergraph.com/resources/)

- [Demos](https://www.tigergraph.com/demo/)
- [Webinar & Events](https://www.tigergraph.com/resources/#webinars&events)
- [Videos & Podcasts](https://www.tigergraph.com/resources/#videos&podcasts)
- [Benchmarks](https://www.tigergraph.com/resources/#benchmarks)
- [Reports & Whitepapers](https://www.tigergraph.com/resources/#reports&whitepapers)
- [Datasheets](https://www.tigergraph.com/resources/#datasheets)
- [Glossary](https://www.tigergraph.com/glossary/)

## Company

### [Company](https://www.tigergraph.com/company/)

- [Company Overview](https://www.tigergraph.com/company/)
- [Leadership](https://www.tigergraph.com/leadership/)
- [Legal terms](https://www.tigergraph.com/terms/)
- [Patents](https://www.tigergraph.com/patents/)
- [Compliance](https://www.tigergraph.com/compliance/)

### [Customers](https://www.tigergraph.com/customers/)

### [Press Room](https://www.tigergraph.com/press-releases/)

### [Partners](https://www.tigergraph.com/partners/)

![white](https://info.tigergraph.com/hubfs/white.svg)

[Privacy Policy](https://www.tigergraph.com/privacy-policy/) [Terms of service](https://www.tigergraph.com/terms/)

Copyright © 2025 TigerGraph. All rights reserved.

[![linkedin](https://info.tigergraph.com/hubfs/linkedin.svg)](https://www.linkedin.com/company/tigergraph/) [![youtube](https://info.tigergraph.com/hubfs/Icon.svg)](https://www.youtube.com/c/TigerGraph)

```json
{
  "@context" : "https://schema.org",
  "@type" : "FAQPage",
  "mainEntity" : [ {
    "@type" : "Question",
    "acceptedAnswer" : {
      "@type" : "Answer",
      "text" : "A knowledge graph gives an LLM structured facts and the relationships between them. Instead of guessing from loosely related text, the model works from a connected view of entities, which improves accuracy on complex questions, reduces hallucination and makes it possible to trace how an answer was formed."
    },
    "name" : "How do knowledge graphs improve LLM accuracy?"
  }, {
    "@type" : "Question",
    "acceptedAnswer" : {
      "@type" : "Answer",
      "text" : "Queries run against the knowledge graph, often alongside vector search, to assemble the entities and relationships relevant to a question. That connected context is passed to the LLM in the prompt, and the model generates an answer grounded in those specific facts rather than in its training data alone."
    },
    "name" : "How do you use a knowledge graph with an LLM?"
  }, {
    "@type" : "Question",
    "acceptedAnswer" : {
      "@type" : "Answer",
      "text" : "A fixed benchmark tests one architecture. Leaving the design open shows which choices teams make on their own when solving real problems. When hundreds of independent builds converge on the same patterns, that is strong evidence those patterns work."
    },
    "name" : "Why run an open experiment instead of a benchmark?"
  } ]
}
```

```json
{
  "@context" : "https://schema.org",
  "@type" : "BreadcrumbList",
  "itemListElement" : [ {
    "@type" : "ListItem",
    "item" : "https://www.tigergraph.com/",
    "name" : "TigerGraph",
    "position" : 1
  }, {
    "@type" : "ListItem",
    "item" : "https://info.tigergraph.com/the-connected-intelligence-experiment",
    "name" : "The Connected Intelligence Experiment",
    "position" : 2
  } ]
}
```