Dify

Open-source LLM app platform

agentslow codeFree tier

Overview

Dify is an open-source platform for building LLM applications. Features visual workflow builder, RAG pipeline, agent tools, and multi-model support. Self-host for free or use cloud. 70K+ GitHub stars. One of the fastest-growing open-source AI projects.

Best For

+LLM app development
+RAG systems
+Custom chatbots
+Enterprise AI

News & Updates

New Open-Source AI-Powered CRM Platform Launches on GitHub with Autonomous Agent Workflows

A new open-source project called CRM has been released on GitHub, positioning itself as an 'agentic-first' customer relationship management platform. Unlike traditional CRMs that rely on manual data entry and human-driven workflows, CRM is designed around AI agents that autonomously handle tasks like lead tracking, follow-ups, and data enrichment. The project is early-stage, with only 4 points on Hacker News and no comments, but its agentic-first approach signals a shift toward AI-native business tools. By making the code open-source, developers can inspect, modify, and self-host the system, potentially disrupting established players like Salesforce and HubSpot. The release comes as enterprises increasingly seek AI-driven automation to reduce manual overhead in sales and support processes.

other8/2/2026

Judge Denies xAI's Bid to Block Minnesota Ban on 'Nudify' Apps

A federal judge has denied xAI's request to block a Minnesota law that bans apps capable of generating non-consensual intimate images, often called 'nudify' apps. The lawsuit, filed by xAI, argued the ban is overly broad and could restrict legitimate AI development. The judge ruled that the state's interest in protecting privacy and preventing harm outweighs xAI's concerns. This decision allows the ban to take effect, setting a precedent for how states can regulate AI-generated content.

ai_society8/1/2026

MindForge Method Enables Small AI Models to Generate Complete Programs Without Source Code

Researchers Yihao Chen, Shi Chang, and Khaled Chawa introduced MindForge, a method that trains small language models to perform whole-life-cycle software engineering—including building complete programs from scratch—using source-free program synthesis. Unlike existing coding agents that only modify existing codebases (e.g., bug fixing or feature implementation), MindForge generates entire programs without requiring access to source code. Evaluated on ProgramBench, frontier models still struggle with this task, but MindForge enables smaller models to achieve competitive results. This approach could democratize advanced software engineering capabilities beyond large, expensive models.

breakthroughs7/30/2026

Researchers Propose Method to Certify Camera Images Against GenAI Tampering

Umar Masud, Abhijith Punnappurath, Luxi Zhao and colleagues have developed a technique to verify whether a photo was captured directly by a camera sensor or altered by generative AI. Their method exploits physical inconsistencies introduced by GenAI, such as unnatural noise patterns or lens distortion signatures that differ from real camera optics. By analyzing these artifacts, the system can distinguish authentic captures from AI-edited or fully synthetic images with high accuracy. This addresses a growing crisis of trust as smartphones increasingly embed GenAI features that can modify photos in-camera, blurring the line between real and generated content.

new_tools7/27/2026