1# Azure Document Intelligence — TypeScript SDK Quick Reference
2 
3> Condensed from **azure-ai-document-intelligence-ts**. Full patterns (custom models, classifiers, batch polling)
4> in the **azure-ai-document-intelligence-ts** plugin skill if installed.
5 
6## Install
7```bash
8npm install @azure-rest/ai-document-intelligence @azure/identity
9```
10 
11## Quick Start
12 
13> **Auth:** `DefaultAzureCredential` is for local development. See [auth-best-practices.md](../auth-best-practices.md) for production patterns.
14 
15```typescript
16import DocumentIntelligence, { isUnexpected, getLongRunningPoller, AnalyzeOperationOutput } from "@azure-rest/ai-document-intelligence";
17const client = DocumentIntelligence(endpoint, new DefaultAzureCredential());
18```
19 
20## Non-Obvious Patterns
21- REST client — `DocumentIntelligence` is a function, not a class
22- Analyze path: `client.path("/documentModels/{modelId}:analyze", "prebuilt-layout").post({...})`
23- Must use `getLongRunningPoller(client, initialResponse)` then `poller.pollUntilDone()`
24- Local file: send as `base64Source` in body, not as binary stream
25- Pagination: `import { paginate } from "@azure-rest/ai-document-intelligence"`
26 
27## Best Practices
281. Use `getLongRunningPoller()` — document analysis is async, always poll
292. Check `isUnexpected()` — type guard for proper error handling
303. Choose the right model — prebuilt when possible, custom for specialized docs
314. Handle confidence scores — set thresholds for your use case
325. Use `paginate()` helper for listing models
336. Prefer neural mode for custom models over template
34