Skip to content
ZHIVEXDocs
Implementation guides

Build your AI layer
with a clear path.

Start with one server-side request. Add streaming, tools, durable sessions, and production controls only when your product needs them.

server / quickstart portable
$bun add @zhivex-ai/sdk @zhivex-ai/openai
›create provider model
›generate, stream, or run an agent
✓normalized result + usage + steps
Choose your runtime

Two SDKs. One adoption model.

Keep application logic portable and opt into provider-native features deliberately.

TypeScript learning path

From your first request to production.

Follow these chapters in order, or jump to the task you need. The sidebar and next-page links follow the same path.

Python learning path

From your first request to production.

Follow these chapters in order, or jump to the task you need. The sidebar and next-page links follow the same path.

Production boundary

The SDK runs inside your application policy.

Zhivex normalizes model and agent behavior. Your server remains responsible for identity, tenancy, credentials, tools, durable storage, billing, and retention.

Review the production guide
01ClientUI + input
02Your serverAuth + policy + SDK
03ProvidersModels + native APIs
Durable stateSessions · runs · workflows · artifacts
Latest releases

See what changed before you upgrade.

Curated impact, affected packages, migration guidance, and links to the complete source changelogs.

Source-backed documentation

These guides are distilled from the public contracts, runnable examples, and production notes in both SDK repositories.

Copied