Dify - LLMOps platform, Define your AI-native apps

Dify.AIAug 4, 2023 3 min read

With the continuous evolution and capability improvements of large language models (LLMs), the demand for AI applications is growing. However, for most developers, building AI applications with GPT and other large language models, frameworks like Langchain, remains an extremely challenging task. Developers have to spend a lot of time learning various abstruse concepts and technical research, and it's hard to continuously operate AI applications.

What if there was a perfect solution: developers don't need to care about the underlying principles of the infrastructure, just focus on turning imagination into real AI applications and operating them continuously, making large language models as easy to use as cloud computing services.

That's why Dify.AI was born! It does it for you.

Now, turning imagination into real AI applications has become easy.

What is Dify.AI?

Dify.AI is an open source LLMOps platform that helps developers build AI applications more simply and quickly. Its core idea is to define various aspects of AI applications, including Prompts, Contexts, and Plugins, through declarative YAML files. Dify provides visual Prompt orchestration, operations, dataset management and more. These enable developers to complete AI app development in days, or quickly integrate LLMs into existing apps, operate and improve them continuously, creating truly valuable AI applications. The hilighted capabilities are:

LLMs support: You can choose capabilities based on different LLMs to develop your AI apps on Dify. Dify is compatible with Langchain, meaning we'll gradually support various LLMs. Currently supported model providers:

OpenAI:GPT4/GPT3.5-turbo/GPT3.5-turbo-16k/text-davinci-003

Azure OpenAI Service

Anthropic:Claude2/Claude-instant

Hugging Face Hub

We provide the following free resources for all registered cloud users (login dify.ai):

1,000 Claude model call credits for building Claude-based AI apps

200 OpenAI model call credits for building OpenAI-based AI apps

Visual Prompt orchestration: Publish an AI app in minutes by writing and debugging prompts visually.
Text Embedding (Datasets): Fully automated text preprocessing, using your data as context without needing to understand abstruse concepts and techniques. Supports PDF, txt and more. Sync data from Notion, web, APIs.
API-based development: Backend as a service. Access web apps directly or integrate APIs into your apps without worrying about complex backend architectures and deployments.
Plugin capabilities: The Dify "Explore-Chat" platform already supports 1st party plugins like web browsing, Google Search, Wikipedia lookup, enabling online search, webpage analysis, showing the AI's reasoning in conversations.
Team Workspaces: Team members can join Workspaces to edit, manage and use team AI apps.
Data annotation and improvement: Visually inspect AI app logs and improve data annotations, observing the AI's reasoning process to continually improve its capabilities.