etrebels/ai-architecture-calculator — explained in plain English
Analysis updated 2026-05-18
Decide whether to ground an AI system with plain document search, a property graph, or a knowledge graph based on your own priorities.
Compare the tradeoffs between AI retrieval approaches on accuracy, cost, latency, and auditability before building a system.
Import a single scoring function into your own app to evaluate architecture choices programmatically.
| etrebels/ai-architecture-calculator | 0xkinno/astraea | 0xkinno/halcyon | |
|---|---|---|---|
| Stars | 0 | 0 | 0 |
| Language | TypeScript | TypeScript | TypeScript |
| Setup difficulty | easy | hard | hard |
| Complexity | 2/5 | 4/5 | 4/5 |
| Audience | developer | developer | developer |
Figures from each repo's GitHub metadata at analysis time.
This project is a small TypeScript library that helps technical decision makers choose how to connect an AI language model to their own data. It compares four common approaches, ranging from letting the AI answer purely from what it already knows, to giving it a searchable set of documents, to two different ways of organizing information as a network of connected facts. Each approach is scored against how accurate it tends to be, how much it costs to run, how fast it responds, and how easy its answers are to audit or verify afterward. Rather than picking one option for you, the tool asks a set of questions meant to surface which of those four factors matters most for your situation, then scores each architecture option against your specific priorities. The scoring itself is not arbitrary. The README states it is based on a combination of published industry benchmarks and academic research papers, including two peer-reviewed papers by the tool's authors, both linked from the project's website. The code itself is framework-agnostic, meaning it does not depend on any particular app framework, and it has no runtime dependencies once installed. A developer can install it with npm, then import a single function along with a set of default parameters to get a scored comparison back. There is also a live, interactive version of the same tool hosted online, where this repository is described as the open version of that model that developers can inspect and adjust. This tool was built by LangOptima, a company that builds AI data systems for other businesses, and is offered as one of their free open tools. It carries an Apache 2.0 license, so it can be freely used, modified, and redistributed, though the README notes that the LangOptima name and branding itself are not included in that license.
A TypeScript tool that scores four ways of connecting an AI language model to your data, based on accuracy, cost, speed, and auditability.
Mainly TypeScript. The stack also includes TypeScript.
Use freely for any purpose, including commercial use, as long as you keep the copyright notice, the company's name and branding are not included in that permission.
Setup difficulty is rated easy, with roughly 5min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Verify against the repo before relying on details.