@loop/ai

AI provider abstraction layer for Loop Health. Provides a unified interface for OpenAI and Anthropic with support for embeddings, PDF extraction, and text chunking.

Installation


pnpm add @loop/ai

Client Creation


import { createAIClient } from '@loop/ai';
 
const client = createAIClient({
  provider: 'openai',
  apiKey: process.env.OPENAI_API_KEY!,
  model: 'gpt-4o',
});

Embeddings


import { createAIClient } from '@loop/ai';
 
const client = createAIClient({ provider: 'openai', apiKey });
 
// Single embedding
const embedding = await client.embed('What is BPC-157?');
 
// Batch embeddings
const embeddings = await client.embedBatch([
  'BPC-157 benefits',
  'TB-500 mechanism',
]);

Text Processing


import { chunkText, extractPlainTextFromLexical } from '@loop/ai';
 
// Chunk long text for embedding
const chunks = chunkText(longDocument, { maxChunkSize: 500, overlap: 50 });
 
// Extract plain text from Payload CMS Lexical rich text
const plainText = extractPlainTextFromLexical(lexicalContent);

Research Paper Ingestion


import { ingestResearchPaper } from '@loop/ai';
 
await ingestResearchPaper({
  title: 'BPC-157 Gastrointestinal Effects',
  content: paperContent,
  metadata: { doi: '10.1234/example' },
});

Used By

@loop/ai is used by applications that need AI and embedding capabilities:

@loop/embeddings-api — OpenAI embedding generation and vector storage
@loop/admin — Knowledge base sync for CMS content