Dify: Develop chatbots and AI workflows without code

Create and customize a basic chat application by selecting models, adjusting parameters, and setting system instructions, then test via debug and preview to optimize prompts for image generation.

Configure variables in a chatbot using short text fields with names limited to letters, numbers, underscores, and a 50 max, then replace prompts at runtime to see rabbit become wizard.

Learn to use paragraph, select, and number variables in Dify to power chatbots: set inputs, prompts, and constraints for long-text summaries, user sentiment, and multiple variations.

Select a model with vision capability, enable multimodal analysis, and configure high or low image resolution, local or URL uploads (e.g., Unsplash), to analyze images and generate tags.

Create intelligent agent applications without code, attaching tools to extend capabilities like internet search, web scraping, and image generation with DALL-E, enabling real-time information and task automation.

Explore text generation apps using a no-code chatbot builder, configure prompts with variables, and test single input-output outputs across formats like text lists, bullet points, and JSON.

Create a chat flow in Nifi by choosing Chat Flow to design multi-turn dialogues with memory using lm, knowledge retrieval, and the answer node.

Explore workflow applications by building template workflows that orchestrate single-turn automation, using input variables and a question classifier to route to IT, sales, or marketing.

Build an article reader using a file upload variable, extract text with a document extractor, and analyze it with an lm node to generate deep questions about the article.

Build and update a knowledge base by uploading documents and structured data, enabling a large language model to query chunks and deliver accurate, contextual responses.

Learn to create a knowledge base in diffy by uploading documents, choosing chunking strategies, enabling embeddings, and configuring vector search for precise retrieval.

Learn to test a knowledge base with retrieval testing in Dify, using vector search, full text search, and hybrid search to locate relevant chunks and assess results.

Learn how to sync from a website using Gina Reader for information extraction, obtain an API key, and convert five pages from docs.defi.ai into embeddings for vector search.

Integrate a knowledge base into Dify chatbots and workflows by adding knowledge sources, using knowledge retrieval to fetch document chunks, and generating responses with an lm model.

Create a chatbot study application that conducts contextual interviews from knowledge databases, generating open-ended questions to elicit detailed responses and prep for job interviews.

Deploy your Dify chatbot with publish and update, run and test remotely, embed on-site via iframe or HTML code, or install a Chrome extension to share a URL.

Track application activity with logs and annotations to monitor conversations and refine responses through user feedback. Create and manage annotations and embeddings to generate similar answers for related questions.

Monitor your chatbot app's performance over time with real-time activity, maintenance controls, and customizable settings. Analyze conversations, active users, token usage, and user satisfaction to supervise performance.

Explore the explore tab to discover pre-created apps like No Tebow by Defy, install the podcast generator, and build a podcast workflow using input, analysis, and model steps.