Getting Started
Projects
Create projects, choose dimensions and metrics, and understand the project lifecycle.
Creating a Project
A project is your isolated workspace in KyroDB. When you create a project in the console, you define a contract that cannot be changed after creation:
| Field | Description | Example |
|---|---|---|
region | Deployment region for your data | us-east-1 |
dimension | Embedding vector dimension | 768, 1536 |
metric | Distance metric for similarity | cosine, dot_product, euclidean |
tier | Service tier | free, pro |
Immutable Contract
Once created, a project's dimension, metric, and region cannot be changed. Create a new project if you need different settings.
Project Lifecycle
After creation, your project progresses through provisioning states before it's ready to accept traffic:
| State | What It Means | What To Do |
|---|---|---|
pending | Project created, not yet provisioned | Wait — provisioning starts automatically |
provisioning | Resources are being allocated | Wait — typically completes in under a minute |
provisioned | Ready to accept traffic | Copy your endpoint from Settings and start sending requests |
failed | Provisioning encountered an error | Check the error in Settings, or contact support |
Choosing Your Dimension
Set the dimension to match your embedding model's output size. Common values:
| Model | Dimension |
|---|---|
| OpenAI text-embedding-3-small | 1536 |
| OpenAI text-embedding-3-large | 3072 |
| Cohere embed-v3 | 1024 |
| Sentence Transformers (all-MiniLM) | 384 |
| BGE / E5 models | 768 |
Choosing Your Metric
cosine— Best for normalized embeddings (most common). Use with OpenAI, Cohere, and most embedding APIs.dot_product— Use when embedding magnitudes carry meaning (e.g., relevance scoring).euclidean— Use for spatial distance comparisons.
When To Create Multiple Projects
Use separate projects when you need different dimensions, metrics, or regions. Use namespaces within a project for logical data separation (e.g., different tenants or environments).