gpt-instagram

Report Abuse

Basic Information

GPT-Instagram is a GPT-based autonomous multi-agent web application built with Next.js, LangChain.js and LangGraph.js that analyzes a user's historical Instagram data via the Instagram Basic Display API to extract personality and posting style and then researches and recommends Instagram post ideas and improvements in response to user queries. The project orchestrates multiple specialized agents to study past posts, condense explanations, generate caption and post suggestions, critique virality potential, create images, and iteratively improve posts with user input. It integrates OpenAI for language models and Tavily for search, stores data in a Postgres database managed via Drizzle ORM, and requires IG access and API keys configured in an environment file. The README includes setup instructions, token guidance, and notes that this was developed as a Tavily AI intern toy project.

Links

Categorization

App Details

Features
The repository provides a multi-agent pipeline composed of a Data Explainer Agent, an Explanation Condenser Agent, an IG Post Generator Agent, a Viral Critique Agent, an Image Generator Agent, and a Post Improver Agent. It is implemented as a Next.js app using LangChain.js and LangGraph.js for agent orchestration and LLM interactions. The app consumes Instagram Basic Display API data using a user access token and integrates OpenAI and Tavily Search APIs. Persistence is handled with Postgres and Drizzle ORM. The repo contains an env.example file describing required keys, setup steps that include yarn install, yarn generate, yarn migrate and yarn dev, and a video walkthrough and README guidance for generating required tokens.
Use Cases
This project helps social media creators and marketers by automating research and ideation for Instagram content based on a user"s historical posts and inferred personality. It analyzes prior posts to surface stylistic patterns, generates caption and image concepts tailored to that style, provides marketing-focused critique to improve potential virality, and supports iterative refinement through a post improver agent. The integrated image generator produces visual concepts that match suggested captions, and the Tavily search integration and OpenAI models augment the research and suggestion process. The tooling reduces manual idea generation and offers structured feedback, while setup notes and examples allow developers to run and adapt the app for experimentation and prototyping.

Please fill the required fields*