What is MediumGPT?
The Building a ChatGPT Plugin for Medium article published on Medium Engineering provides a technical exploration of the capabilities and proof of concept of ChatGPT plugins. OpenAI has recently introduced plugins for ChatGPT to enable access to up-to-date content on the internet. ChatGPT had been trained on a 2021 dataset and had limited access to the internet, however, plugins can solve this problem by allowing platforms such as Medium to expose their content to ChatGPT. The user prompt and installed plugins enable ChatGPT to trigger the correct API to retrieve a piece of content. Chat GPT plugin support is currently in private alpha, with access granted to Medium to explore the possibilities of making their content retrievable by ChatGPT. The article explains how to build a ChatGPT plugin with a well-known/ai-plugin.json file, which provides necessary information so that ChatGPT can understand that your domain supports plugins and link the appropriate description, icons, and links to the end-users when installing the plugin. The OpenAPI YAML file exposes the schema that allows ChatGPT to understand the API, and the Go language can be used to write the MediumGPT service that converts hardcoded GraphQL queries into REST APIs. The article focuses on the technical side of building a ChatGPT plugin and demonstrates how it works.
Pros
- Supports plugins
- Accesses up-to-date content
- Triggers correct API for content
- Alpha access
- Capable of complex queries
- Ability to chain queries
- Supports Go language
- Solves lack of internet access
- Understands API surface
- Enables browsing and manipulation
- Ability to expose APIs
- OpenAPI YAML for schema
- Handles both authenticated and non-authenticated users
- Proof of concept available
- Supports platform specific plugins
- Allows content exposure to ChatGPT
- Can convert GraphQL queries into REST APIs
- Supports OpenAPI specification
- Content retrieval from Medium
- Individual post content retrieval
- Allows searching posts with specific query
- Provides descriptions of schema objects
- Ensures context-aware responses
- Plugin debugging through ChatGPT interface
- Supports middleware service creation
- Understands complex descriptions
- Enables natural query language
- Allows adjustment to API description
- Allows debugging directly on ChatGPT
- Provides real-time debugging feedback
- Recognizes trending posts
- Searches posts using specific queries
- Fetches post content
- Retrieves relevant post summaries
- Offers browsing and manipulation for Medium content
- Allows detailed technical exploration
- Allows complex and useful queries
- Handles plugins in private alpha
- Retrieves related posts to a topic
- Allows technical demonstrations
- Enables automated response generation
- Supports chained API calls
- Responsive to user prompts
- Capable of query manipulations
- Supports automated content retrieval
- Allows post condition querying (e.g.
- paywall status)
- Provides results as simple
- clear JSON results
Cons
- In private alpha
- Requires domain support
- Dependent on well-documented schemas
- Limited plugin support
- API access depends on user prompts
- Content retrieval issues
- Cannot directly talk to GraphQL API
- Limited service languages (Go)
- API chaining complexity
- Potential confusion with multiple plugins
MediumGPT FAQ
What is MediumGPT?
MediumGPT is a ChatGPT plugin designed for the Medium platform. It's a feature created by OpenAI that allows Medium to expose its content to ChatGPT, effectively making Medium content retrievable by the artificial intelligence system.
What is the purpose of the ChatGPT plugin for Medium?
The purpose of the ChatGPT plugin for Medium is to enable more dynamic interactions between users and the Medium content. It aids in solving a significant issue of ChatGPT, which is the AI's limited access to the internet and the inability to access up-to-date content. Through the ChatGPT plugin, Medium can expose its content to ChatGPT, which can then retrieve relevant content based on user prompts and installed plugins.
How does the ChatGPT plugin for Medium work?
The ChatGPT plugin for Medium works by creating an interface between the Medium's data and the ChatGPT model. MediumGPT service, written in the Go language, interfaces with Medium's GraphQL API and exposes parts of it as a REST API in a format that ChatGPT can understand. ChatGPT uses an OpenAPI YAML file to interact with the API and understand the available services, triggering appropriate requests based on user prompts.
What problems does the ChatGPT plugin solve for Medium?
The ChatGPT plugin for Medium solves one of ChatGPT's significant issues – its lack of direct internet access and up-to-date content. Before the introduction of plugins, ChatGPT was trained on a dataset from 2021, and didn't have access to the internet. With the plugin, Medium can expose its content to ChatGPT, ensuring the content retrieved is current and relevant.
What is the current status of the ChatGPT plugin development?
The ChatGPT plugin development is currently in the private alpha stage. Medium has been granted access to explore the possibilities of making their content retrievable by ChatGPT.
How do I build a ChatGPT plugin?
Building a ChatGPT plugin involves creating an ai-plugin.json file on your domain. This file provides necessary information so that ChatGPT can identify the domain's plugin support. It links appropriate descriptions, icons, and links to the user when they are installing the plugin. A key element is the OpenAPI YAML file that exposes the schema allowing ChatGPT to understand the API.
What is the use of the ai-plugin.json file in the ChatGPT plugin?
The ai-plugin.json file in the ChatGPT plugin provides necessary information so that ChatGPT can recognize that your domain supports plugins. It links the suitable description, icons, and links to end-users when they install the plugin. It is pivotal in helping ChatGPT understand your API services and to interact with them accordingly.
What programming languages can be used to write MediumGPT services?
The MediumGPT service that converts hardcoded GraphQL queries into REST APIs can be written using the Go programming language.
