Loading pattern preview
BlocksTools & Utilities
Local RAG Chat
Local RAG Chat
Load a sample Local RAG index, rank chunks in the browser, and stream a cited answer without a vector database.
Use this when you already have an exported embedding index of text chunks and want users to chat over it with cited, streamed answers.
Loading pattern preview
Problems solved
Technical challenges this implementation handles out of the box.
- Import a portable embedding index and chat over it in the browser
- Rank local vectors with AI SDK Core cosineSimilarity
- Stream cited answers from bounded selected text without a vector database
Use cases
Products and workflows this pattern is designed to support.
- Query a portable documentation index without a database
- Local RAG demos and workshops
- Cited chat over an exported embedding JSON file
Compare with
Similar patterns, and when each one is the better fit.
Local RAG IngestUse this when a small docs or knowledge-base prototype needs its text turned into embeddings held in React state and saved as a portable index.
Local Image SearchUse this when you already have an exported image index and need to search it by meaning in the browser using caption embeddings.
Local Image IngestUse this when teaching or prototyping caption-then-embed for images and you want a downloadable index file as the result.
Jev AEO AuditUse this when auditing a page for AEO and you want title, canonical, and noindex checks in code while Jev judges the citation-worthiness.
Setup
Requirements, wiring steps, and what this pattern adds to your project.
Getting started
Pick how you want to pull this pattern in. Then wire env vars and routes the same way.
- 1
Download → Next.js app
Use Download → Next.js app in the preview toolbar. You get a zip scaffold (Pro downloads rewrite preview `/view/.../api/*` paths to `/api/*` automatically).
- 2
Extract the zip and install dependencies
Unzip the archive, then run `pnpm install` (or npm / yarn) in the project folder.
- 3
Fill .env.local
The download includes .env.local placeholders when needed. Add real values for: AI_GATEWAY_API_KEY.
- 4
Customize the agent and tool files
Adapt prompts, tools, and stop conditions in the generated project for your product.
- 5
Run the app and open the pattern route
Run `pnpm dev`, open the setup/checklist page if present, then navigate to the generated pattern route and verify responses.
Environment variables
AI_GATEWAY_API_KEYGet key
Capabilities
AI SDK APIs
embedcosineSimilaritystreamTextsafeValidateUIMessagesconvertToModelMessagescreateUIMessageStreamtoUIMessageStreamcreateUIMessageStreamResponsegatewayExternal services
gatewayvercel-ai-gatewayUpstash
Files added
29 files
app/page.tsxapp/layout.tsxapp/api/query-embedding/route.tsapp/api/chat/route.tscomponents/local-rag-chat.tsxcomponents/index-import.tsxcomponents/local-citations.tsxhooks/use-local-rag-chat.tslib/chat-contract.tslib/chunk-text.tslib/embedding-config.tslib/local-rag-index.tslib/rate-limit.tslib/request-utils.tslib/retrieval.tslib/sample-indexes.tsfixtures/acme-ops-handbook.jsonfixtures/cedar-product-faq.jsonfixtures/atlas-deploy-runbook.jsonfixtures/sources/pto-and-holidays.mdfixtures/sources/tools-and-access.mdfixtures/sources/expense-policy.mdfixtures/sources/cedar/sharing-and-permissions.mdfixtures/sources/cedar/plans-and-billing.mdfixtures/sources/cedar/export-formats.mdfixtures/sources/atlas/rollback-procedure.mdfixtures/sources/atlas/on-call-and-paging.mdfixtures/sources/atlas/feature-flags.mdREADME.md
Dependencies
11 total
npm packages
7AISDKUpstashlucide-reactNext.jsReactstreamdownZod
Registry components
badgebuttontextareatooltip
Critical files
4 files
lib/local-rag-index.tsCopied V1 portable embedding-index contract. Must stay parity-tested against local-rag-ingest. Reject incompatible schema/version/model/dimensions/chunking before cosineSimilarity.
lib/retrieval.tsBrowser-side ranking with AI SDK Core cosineSimilarity, threshold/top-K/context budget selection, and citation helpers. Keep ranking in the browser; never upload the full index to rank on the server.
query-embedding/route.tsStateless Core embed boundary for one query string. Locked model openai/text-embedding-3-small with providerOptions.openai.dimensions=512. Never add @ai-sdk/openai solely for embeddings.
chat/route.tsCited answer boundary: validate retrieval body without embeddings, assign S1..S5 server-side, write non-transient data-retrieval, convert prior data parts to undefined, streamText via gateway deepseek/deepseek-v4.1-flash.
Requirements
- Complexity
- Intermediate
- Setup time
- ~10 minutes
- Node
- 24+
- Memory
- 256MB
Upstash Redis for deployed demo rate limitingA completed local-rag-ingest JSON export
