Dify

Thrilled to announce Dify's successful SOC 2 compliance! Their dedication to security and transparency shines through. With the SOC 2 report, they're proving their robust controls and processes, solidifying trust and credibility. Hats off to the team! #SOC2 #Compliance #Security #Trust

?? Dify is now officially SOC II certified. We are consistent compliant with security and data safety from Day 1, and now we are excited to announce Dify is AICPA SOC II Certified. You can now build your own App on Dify assuring confidentiality and privacy fine-grain handled by our system.

??? Dify v0.9.2 is out! We’ve got some new integrations for you to explore: ?? New Integrations 1?? Qwen2.5-72b and Llama3.2 in OpenRouter 2?? O1-mini, O1-preview, and LlamaGuard in Azure 3?? Discord (Incoming Webhook): Allows AI apps to send messages to Discord automatically. ?? Learn how to set it up here: https://lnkd.in/gTR_V_SK 4?? Volcengine VikingDB: A new vector database provider stores, indexes, and manages massive embedding vectors ?? New System Variables sys.app_id, sys.workflow_id, and sys.workflow_run_id These system variables identify AI workflows for monitoring and data analysis. ?? Many thanks to the community for these contributions. See the full changelog here: https://lnkd.in/gGHsiuRz

We’re excited to be part of the October MLOps Now Meetup! Our frontend developer YI XIAO will be talking about “Enhancing AI Workflows with Parallel Processing and Custom LLM Agents”. Join us on October 25th to learn about the latest in AI Agents, RAG, and LLM development. Don’t miss out! ?? Sign up here: https://bit.ly/3Yj5p9E

?? MLOps Now 10? ?? ?? - Meet Agent & LLM Builders VESSL AI? ????? LLMOps ????? Dify? ???? 10? ??? ???? ????? <LLM in Production>? ??? ?? AI Agent ???, RAG ? LLM ??? ???? ?? ??? ??? ????? ??? ?????. ???? ???? https://bit.ly/3Yj5p9E - ??: 2024? 10? 25? (?) ?? 2?-5? 30? - ??: ???? ????? &Space (https://naver.me/xbwWoebp) ??? ?? ?? 1/ From AI Agents to AGI: A Simple Guide?? ??- Jaeman An, Co-founder & CEO, VESSL AI 2/ Enhancing AI Workflows with Parallel Processing and Custom LLM Agents (*This session will be delivered in English)?? ??- YI XIAO , Frontend Developer, Dify 3/ ???x??????? ???? ???? LLMOps ??? - ???, ML Engineer, ???? 4/ ??? LLM workload? ?? ??? serving configuration ?? - ??? (Hyungjun Kim), CEO, SqueezeBits 5/ ??? RAG, ??? ???? ???? ?? ???? ??? ??? - ??? (Jeffrey Kim), Co-founder, AutoRAG - ?? October MLOps Now Meetup with LLMOps Startup Dify - Meet Agent & LLM Builders Presentations will be on the topic of <LLM in Production>, covering the latest trends in AI Agents, RAG, and other key techniques applied to LLM development. Sign up for the event??https://bit.ly/3Yj5p9E - Date: Friday, October 25, 2024, 2:00 PM - 5:30 PM - Venue: Startup Alliance &Space (https://naver.me/xbwWoebp)

??? Dify v0.9.1-fix1 has been released. This update fixes the issue where conversation history was lost during API calls to AI applications, causing multi-turn conversations to fail. ?? Please note that the upgrade and data migration may take some time if you have a large amount of historical conversation data. See the full changelog here: https://lnkd.in/gSZKk7ns

?? Dify v0.9.0 is out! ? We’ve upgraded our knowledge base both internally and externally to provide you with more options and greater accuracy.?? ?? External Knowledge Base? We’ve made connecting external knowledge base easier than ever, giving you more options. By setting up an API endpoint, API key, and external knowledge ID, your AI applications can now retrieve relevant information chunks from external sources based on user queries. Get started with our API reference and tutorial: https://lnkd.in/g4H7ZWGe ?? New Integrations -Jina Reader: When creating an internal knowledge base, Jina Reader functions as a website data crawler, converting entire sites into clean, structured data, making it perfect for LLM apps. Exclusively available for Dify users, try it out! Cookbook: https://lnkd.in/g__Uxhnw?? - Llama 3.2: You can now use Llama 3.2 to build AI apps in this version provided by Fireworks and Groq, enabling deep image understanding and reasoning.?? For more new models and enhancements, check out our changelog: https://lnkd.in/g4KCP4iS

Catch Dify at #SFTechWeek! We’ll be at HackNight, hosted by Weaviate at GitHub.?? ?? Evening of Fun: - Lightning Talks (Weaviate, Dify, Wordware (YC S24), FriendliAI, APARAVI) - Demo Challenge: Showcase your creativity by combining Quickstarts or technologies. - Quickstart Challenges - Social & Network Challenges?? ?? Oct 8, 16:00 - 21:00 GMT-7? ?? San Francisco, California? ?? Sign up here: https://lnkd.in/e7Qt4XiJ

Community Tutorial ?????? Check out this great tutorial video on how to install and set up a self-hosted Dify environment on Google Cloud VM for more flexibility and better data security. You’ll also learn how to create and run Dify workflows, and how to use the Dify API to trigger workflows and integrate them smoothly into your apps. Watch the full video here: https://lnkd.in/g2Umc_Fb

We’re excited to announce the Qwen 2.5 series model is now live on Dify v0.8.3! ?? ?? With multilingual capabilities, the Qwen 2.5 model handles up to 128K tokens in context and generates outputs of up to 8K tokens, greatly improving performance with long-form content. It also offers notable improvements in math and programming tasks. Give it a try! ?? Plus, we’ve introduced a timeout configuration in v0.8.3. Developers can now set custom time limits for text generation and workflows by adjusting the `TEXT_GENERATION_TIMEOUT_MS` variable in the `docker/.env.example` file. This is especially useful when you need precise control over task execution times. See the full changelog here: https://lnkd.in/gr5cj_kY