The term "Dify" is derived from Define + Modify, which means to define and continuously improve your AI application.
With the emergence of various capabilities of large language models (LLMs), the scenarios for AI applications have become much broader. However, for most developers, developing AI applications based on large language models such as GPT and technology frameworks like Langchain is still a challenging task. Developers must spend a lot of time learning various obscure concepts and technical research, and they cannot carry out continuous operations of AI applications.
Is there a tool that allows developers to focus on transforming their imagination into practical AI applications without worrying about the underlying principles of infrastructure? Can it make large language models as easy to use as cloud computing services?
Based on this question, Dify.AI was born.
Do it for you
What is Dify.AI?
Dify.AI is an LLMOps platform that helps developers build AI applications more simply and quickly. Its core concept is to define various aspects of AI applications through declarative YAML files, including prompts, context, and plugins. Dify provides visual prompt orchestration, operation, dataset management, and other functions. These features enable developers to complete the development of AI applications within days or quickly integrate LLMs into existing applications while continuously operating and improving them to create truly valuable AI applications.
Currently supported capabilities:
01. Visual Prompt Orchestration
Create prompts through a visual interface and debug them in just a few minutes to publish an AI application.02. Long Context Integration (Dataset)
Automatically preprocess text using your data as context without understanding obscure concepts or technical processing.03. API-based Development
Backend-as-a-service. Access web apps directly or integrate APIs into your application without worrying about complex backend architecture or deployment processes.04. Data Annotation and Improvement
Visually review AI logs and improve data annotation, observe AI's inference process, and continuously enhance its performance.
In just four steps, you can create AI applications that adapt to various scenarios, such as personalized chatbots, AI-powered customer service, and text generation based on professional knowledge.
Currently, we have completed support for the OpenAI GPT series and are integrating Azure OpenAI Service, Claude, Hugging Face HubHugging Face Hub models. Plugin capabilities will be supported soon to unleash the unlimited potential of large language models.
Where does Dify.AI come from?
Dify.AI is built by core members of the original Tencent Cloud CODING DevOps team. We have been deeply involved in the SaaS developer tools product field for many years. We have successfully created several collaborative developer products and made CODING a one-stop DevOps platform with millions of developers in China. These experiences have greatly helped us create Dify, an LLMOps product.
If you are interested in Dify products and team, please feel free to contact us: hello@dify.ai
Launched for 36 hours
On May 9, 2023, Dify.AI was officially launched in the production environment. Before a small-scale public beta test could be conducted, the founder of the team spontaneously posted an update on a community platform, which led to widespread sharing by users. In just 36 hours, over 1,500 applications were created and the open-source project received more than 240 stars (even though it hadn't been officially open-sourced yet). At the same time, we received numerous positive reviews and acknowledgments from users. We are grateful for the encouragement from our community friends.
Open Source for Natural Emergence
We believe that when AI applications are ubiquitous, they will generate unimaginable influence and value, creating natural emergence.
We are about to open-source Dify so that developers worldwide can participate. We aim to build an open ecosystem by leveraging Dify's open source and continuous development to promote large language model adoption and help innovators create better AI applications.
via @dify_ai and @goocarlos
