Dify

Dify x Arklex: testing AI agents before they reach production. We tested the @dify_ai and @ArklexAI integration, which connects ArkSim, Arklex’s open-source agent testing framework, to Dify applications through a lightweight Chat API adapter. Dify handles workflow design, RAG pipelines, tools, and deployment. ArkSim runs realistic multi-turn synthetic users against the Dify app, helping teams uncover hallucinations, context loss, contradictions, and workflow failures before real users encounter them. It also supports evaluation metrics such as helpfulness, faithfulness, coherence, and goal completion, making it useful for CI quality gates and knowledge base regression testing. Read the full walkthrough: dify.ai/blog/dify-x-ar…

🙋 Getting stuck somewhere with AI on your website? Drop it in the comments — we'll make sure to cover it in the session. No question is too basic. Sign up here and let us know 👇

Enterprise AI implementation is never straightforward. In the first session of CIO Taiwan Academy, Dify (LangGenius K.K.) Head of Taiwan Alvin Wu highlights: Only 48% of AI projects globally reach production. Enterprises face three critical challenges: tech fragmentation, talent gaps, and governance blind spots. How does Dify help overcome these barriers? Visual workflows, end-to-end lifecycle management, enterprise security certifications, and 200+ LLM integrations—already validated by 280+ enterprises and government agencies. 📖 Full article: lnkd.in/gXRMykm5 #Dify #GenerativeAI #EnterpriseAI

Last day at #ATxSG2026 today! Huge thanks to everyone who stopped by the past two days ❤️ Our booth has been absolutely packed & buzzing with energy! Such a blast talking about AI initiatives. Today is the final day. Come meet the team! 📍Booth 3K2-1 | Hall 3

OrcaRouter🐳 x @dify_ai💥 Build production-ready agentic workflows with: • 200+ models through one unified API • Adaptive routing strategies (latency, cost, quality, coding-aware) • Full LLM observability across every request/input/output • Built-in guardrails & policy enforcement Instead of wiring together multiple providers manually, route everything from one layer. Use frontier models when needed. OSS models when they’re enough. Monitor every token. Prompt-aware routing for production AI systems. orcarouter.ai

Dify Creator Center and Template Marketplace are now live. Creators can publish and manage workflow templates in one place, while users can easily discover, preview, and reuse them. With Creator Center, you can build your profile, publish templates, and keep them up to date. With Template Marketplace, users can browse by category, tag, or creator to get started faster. Explore and publish: - Creator Center: dify.ai/creator-program - Template Marketplace: marketplace.dify.ai - Learn more: dify.ai/blog/build-sha…

Dify is heading to ATxSG 2026 in Singapore! 🇸🇬 Join us at Singapore EXPO Halls 3, Booth 3K2-1 from May 20 to 22. Come chat with us about your enterprise use cases. Book your 1 on 1 session now! 👇 app.asiatechxsg.com/event/asia-tec… #ATxSG2026 #Dify #AI #LLMOps

Most websites can't answer a question at 2am. Yours could. We're hosting a free live webinar where we'll build an AI chatbot and embed it on a website. You will learn: - Build a chatbot from scratch in Dify - Add knowledge and context for more accurate answers - Test it before it goes live - Embed it on your website Bonus: Learn how to connect it to an external knowledge base. Register to get the full live video and DSL file, so you can import it into Dify and customize it. 👉 luma.com/yphihg1m See you soon! #Dify #AIChatbot #WebsiteChatbot #Agents #CustomerSupport #FridayBuilds #WeekendProject #ShipYourIdeas

🔥 Dify v1.14.1 is out. It’s been a while since the last release, and during that time we’ve been listening closely to community feedback, refining details, fixing rough edges, and rethinking how workflows should actually work in real teams. This release is less about adding more things onto the canvas, and more about making workflows collaborative, reusable, and easier to integrate into existing systems. What ships: - Collaboration Mode: Multiple teammates can now edit the same workflow in real time, with comments, @ mentions, email notifications, and live collaboration helping business, product, and engineering align directly on the canvas itself. - Copy & paste nodes anywhere: Every node (except user input) is now reusable across workflows and even across Dify instances. Proven workflow logic no longer needs to be rebuilt from scratch every time. - HITL Service API: Human input steps can now integrate with enterprise IMs, or internal systems through API access, instead of being limited to the Dify Console. - Stability Improvements: A focused round of fixes across MCP, OAuth flows, plugin upgrades, Marketplace interactions, and overall runtime stability. Read the blog: dify.ai/blog/dify-1.14… Full release notes: github.com/langgenius/dif… Docs: docs.dify.ai/en/use-dify/bu…

Building an AI workflow that decides when to keep researching on its own. ⚡ Currently experimenting with recursive search loops, structured outputs, and autonomous decision-making inside Dify. Still debugging. Still building. 🚀 #AI #Automation #BuildInPublic #Dify #AIAgents

Final call! Join us at AWS Builder Loft in San Francisco tomorrow, May 7, for an evening with @awscloud , Dify, and @PingCAP on what it really takes to move AI agents and GenAI apps from prototype to production. If you’re working on AI agents, production GenAI workflows, or enterprise AI apps, join us to connect with other builders and learn what it takes to ship reliable, scalable AI systems in the real world. See you there! 📍 AWS Builder Loft 525 Market St, 2nd Floor, San Francisco, CA 📅 Thursday, May 7, 2026 🕔 5:00 PM – 8:30 PM Register here: luma.com/v9no5ytc

LangGenius K.K.’s official partner Open Network System (ONS) is at AI EXPO Korea on May 6-8, showcasing and introducing Dify! If you're attending the expo, don’t miss the chance to stop by the booth, see live demos about Dify, and chat with the team. 🚀 #AIEXPOKorea #Dify #AI #ONS

Join us at AWS Builder Loft in San Francisco on May 7 for an evening with @awscloud , Dify, and @PingCAP on what it really takes to move AI agents and GenAI apps from prototype to production. In collaboration with @PingCAP , we’ll be bringing together Bay Area AI app developers, builders, and teams working on scalable GenAI systems for practical talks, demos, and networking. If you’re building AI agents, production GenAI workflows, or enterprise AI applications, this is a great chance to connect with other builders and learn what it takes to ship systems that are reliable, scalable, and ready for real-world use. 📍 AWS Builder Loft 525 Market St, 2nd Floor, San Francisco, CA 📅 Thursday, May 7, 2026 🕔 5:00 PM – 8:30 PM Register here: luma.com/v9no5ytc

On 4/16, Alvin (Head of Middle East & Taiwan) and Kelvin (APAC Solution Architect) from Dify went hands-on with NYU Abu Dhabi MBA students, faculty, and NYU Stern alumni. One question drove the session: what does an AI agent look like when it's shipping work every day? Three things we actually showed (not just talked about): • A scheduled financial analyst agent — runs hourly, aggregates web topics, passes them to a node set up as a senior Abu Dhabi analyst, and routes output through email-based human approval before it ships. Every node logged, every answer traceable. • Dify’s new Human-in-the-Loop node — the missing piece most enterprises need to move past PoC. Critical decisions pause for review; everything else keeps running. • The Knowledge Pipeline — ETL preprocessing + LM-driven Q&A generation + per-chunk citations. The gap between “works in a demo” and “holds up in production.” Alvin also unpacked why ~80% of AI projects never escape pilot — and showed a manufacturing client that built 200+ apps, kept 60 in production after a month, and cut analysis time from 8h to 3h. Thanks to the AI and Tech Society at Stern at NYUAD MBA organizing team!

Munich didn't feel like what we'd imagined a hackathon city would feel like. There's a particular kind of intensity you get when 500+ builders share a hall for 48 hours, sleeping bags on the floor by 3am, and the rest of the room still shipping. @TUMai_official Makeathon was one of the largest we've ever been part of, and honestly the most alive build energy we've felt in Europe. Across the floor, Dify became the orchestration layer for a lot of those projects, the brain routing intent, coordinating models, wiring agents into real systems. Whether someone was an open-source dev shipping something scrappy or a team already aiming at production, the same thing held and that's exactly what we built Dify for. From a weekend prototype to production-ready apps, Dify carries the weight. 😃 Huge thanks to @TUMai_official, and to the partners who made this weekend what it was @Google, @AMD, @cognee_, @ElevenLabs, @osapiens_, Reply, Spherecast (YC S24), @HappyRobot. It genuinely meant a lot to build alongside all of you. Excited to keep building with the European community! #DifyEurope #Dify #DataSovereignty #EnterpriseAI 📷 Photos by Jiachen Zhong

You've built your AI agent on Dify. The drag-and-drop workflow is smooth, the RAG pipeline is solid. But how do you know it actually holds up when real users start pushing it in unexpected directions? That's where ArkSim comes in. Dify makes it fast to build and deploy production-ready agentic workflows. ArkSim closes the gap between "it works in testing" and "it works in the wild" automatically simulating thousands of real-world conversations to surface failures before your users do. Build fast. Ship with confidence. Try Arksim on your next Dify project: github.com/arklexai/arksi… #Dify #AIAgent #AgentEval #ArklexAI #OpenSource

On 4/14, Alvin (Head of Middle East & Taiwan) and Kelvin Tsim (APAC Solution Architect) from Dify were invited by CUQ Ulster University in Qatar for a guest lecture on agentic AI — and the energy was incredible. Here’s what the session covered: - The AI Maturity Roadmap (L0–L5) — From zero AI adoption to fully autonomous digital employees. Alvin focused on the practical sweet spot: L1–L3 deployments that integrate into existing enterprise systems. - The “PoC Trap” — Why most AI pilots never make it to production. Alvin broke down the three core adoption gaps: technical fragmentation, talent mismatch, and governance blind spots. - Real Enterprise Results — A manufacturing client built 200+ AI applications with Dify, kept 60 active in production after just one month, and cut analysis time by ~60% (from 8 hours to 3). That’s a 3x capacity increase: freeing teams to focus on customer-facing work. - Live Platform Demo — Kelvin showed how to build a knowledge-driven chatbot from scratch in minutes — from document upload and chunking to publishing a fully functional agent with retrieval citations. - Arabic & Regional Use Cases — The session also explored Arabic language support, Middle East government deployments for translation and training, and a Malaysian government use case for public Islamic content search. Thank you to CUQ Ulster University for the invitation and the fantastic engagement!

Dify now supports Claude Opus 4.7. Plug it into your Dify workflows and get stronger coding support, sharper image understanding, and agentic tasks that run longer with less hand-holding. Update your Anthropic plugin and give it a try: marketplace.dify.ai/plugin/langgen… #Claude #Anthropic #Dify #LLM #Agents