Feedple

Features

Product

Integrations

Pricing

Docs

FAQ

Client Login

Get Started

Feedple Documentation

Complete technical guides, API references, configuration parameters, and web widget integration guides.

Why not just use ChatGPT, Claude or Gemini?

Feedple AI is an AI execution layer for business systems, while ChatGPT, Claude or Gemini are primarily AI conversation interfaces.

1. Connect AI to your business in minutes

ChatGPT

Connecting ChatGPT directly to your business data typically requires building an MCP server or developing a custom integration using the OpenAI API. This often involves implementing:

REST or GraphQL API integrations
Function calling / tool calling endpoints
Authentication (OAuth, API keys, JWT, etc.)
Database access layer or ORM
Vector database and RAG pipeline (if using knowledge retrieval)
Webhook handling
Permission and access control
Infrastructure to host and maintain the integration

This additional engineering work is required before ChatGPT, Claude or Gemini can securely access your business data.

Feedple AI

Install our SDK, add your API key, and you're connected. No MCP server. No custom integration. No database URL exposure. No need to expose your application's API endpoints. As long as your application is running, Feedple AI can securely access only the data you've explicitly allowed through the SDK.

2. AI that works even when you're offline

ChatGPT

ChatGPT responds when you ask it something. It does not continuously monitor your business or reliably execute business workflows on a schedule.

Example:

"Every day, find inactive customers and send personalized re-engagement emails."

This isn't what ChatGPT is designed to do as a business automation platform.

Feedple AI

Feedple runs automations continuously in the background.

Examples:

Detect inactive customers
Generate personalized emails
Send notifications
Monitor business metrics
Execute scheduled workflows automatically

3. Built for secure business data

Businesses shouldn't give an AI unrestricted SQL access.

Feedple AI introduces a safer execution model:

The AI generates an Intermediate Representation (IR), not raw SQL.
Feedple validates the IR against your permissions and policies.
Only approved operations are compiled into SQL and executed.
Every request is tenant-scoped and auditable.

This provides much stronger control than allowing an LLM to directly generate and execute SQL.

4. Bring AI directly into your product

Your team shouldn't need to open another website every time they want AI assistance.

Feedple provides an embeddable web widget that integrates directly into your admin dashboard, allowing employees to interact with AI where they already work.

Examples:

"Which customers are likely to churn?"
"Show today's failed payments."
"Generate a retention campaign."

No context switching.

5. AI that understands your business

General-purpose AI knows general knowledge.

Feedple learns your business.

By connecting your application data, Feedple builds business context including:

Customers
Orders
Products
Transactions
Support history
Internal workflows
Custom business entities

This enables answers and actions that are grounded in your company's data instead of generic assumptions.

The core difference

ChatGPT is an AI assistant.

You ask questions.

It answers.

Feedple AI is an AI operating layer for your business.

It understands your business, continuously monitors it, and securely takes action on your behalf.

Feedple TypeScript / Node.js SDK

Connect your database to Feedple AI securely through our automated background integration.

Package Information & Resources

Information	Details
Package Name	feedple-sdk
Current Version	`1.0.0` (Latest Release)
Package Manager	npm / yarn / pnpm
Official Registry Link	npmjs.com/package/feedple-sdk

Quick Installation

Install the official feedple-sdk package from npm:

BASH

npm install feedple-sdk

Supported database drivers: pg (PostgreSQL), mysql2 (MySQL), better-sqlite3 (SQLite). Requires Node.js 18.0+.

Quick Start Guide

Initialize Feedple SDK in 3 steps:

PostgreSQL Example

TYPESCRIPT

import { FeedpleSDK, PgAdapter } from "feedple-sdk";
import { Client } from "pg";

// 1. Connect to your database
const pgClient = new Client({ connectionString: process.env.DATABASE_URL });
await pgClient.connect();

// 2. Set table access permissions & Start Feedple SDK
const sdk = new FeedpleSDK({
  apiKey: "sk_live_your_api_key",
  db: new PgAdapter(pgClient),
  identity: {
    allowedTables: ["users", "orders", "products"],
  },
  autoSync: true,
});

await sdk.connect();

MySQL Example

TYPESCRIPT

import { FeedpleSDK, MysqlAdapter } from "feedple-sdk";
import mysql from "mysql2/promise";

const mysqlPool = mysql.createPool(process.env.DATABASE_URL);

const sdk = new FeedpleSDK({
  apiKey: "sk_live_your_api_key",
  db: new MysqlAdapter(mysqlPool),
  identity: { allTables: true },
});

await sdk.connect();

SQLite Example

TYPESCRIPT

import { FeedpleSDK, SqliteAdapter } from "feedple-sdk";
import Database from "better-sqlite3";

const sqliteDb = new Database("app.db");

const sdk = new FeedpleSDK({
  apiKey: "sk_live_your_api_key",
  db: new SqliteAdapter(sqliteDb),
  identity: { allowedTables: ["orders", "items"] },
});

await sdk.connect();

Configuration Options

Option	Type	Default	Description
`apiKey`	`string`	required	Your Feedple workspace API key
`db`	`FeedpleDbAdapter`	required	Database configuration instance (`PgAdapter`, `MysqlAdapter`, or `SqliteAdapter`)
`identity`	`Identity`	required	Table access rules
`autoSync`	`boolean`	`true`	Automatically sync schema changes
`syncInterval`	`number`	`60`	Sync frequency in seconds

Security & Permissions

Control which tables Feedple AI can query:

TYPESCRIPT

// Whitelist specific tables (Recommended)
const identity = {
  allowedTables: ["orders", "products"],
};

// Allow access to all tables
const identity = {
  allTables: true,
};

Note

Sensitive columns like `password`, `token`, `secret`, `hash`, `salt`, `ssn`, and `credit_card` are automatically excluded during schema sync.

Stop the SDK

Shut down the background worker process cleanly when your application exits:

TYPESCRIPT

await sdk.close();

Framework Examples

Explore ready-to-run starter projects from the official feedple-examples repository:

Framework	GitHub Source
Next.js 14 (App Router)	View Code
Express.js	View Code
NestJS	View Code

Next.js 14 (App Router)

Full-Stack Integration · Next.js 14 App Router running the Feedple SDK background worker alongside PostgreSQL schema synchronization. Open Next.js Example on GitHub

Express.js

REST Server Integration · Complete Express REST server demonstrating database schema extraction, SHA-256 change detection, and background WebSocket handler execution. Open Express Example on GitHub

NestJS

Enterprise Service Module · Enterprise NestJS application managing the Feedple SDK background process via OnModuleInit and OnModuleDestroy lifecycle hooks. Open NestJS Example on GitHub