This week's roundup highlights two insightful talks on how open-source tooling is reshaping the way AI agents and machine learning systems are built and improved. From the architecture of agent harnesses to the mechanics of continual learning, these curated videos offer practical lessons for developers and researchers alike.
First, Hugging Face presents a deep dive into DeepSeek Harness (DSH), DeepSeek's open-source agent harness built on the Cordis plugin system. The video explains the project's guiding philosophy—everything is a plugin, including the agent loop, the tools, and even the system prompt—and offers a quick tour of what's new in the framework. It's a compelling look at how modular design can make agent infrastructure more flexible and extensible.
Next, the PyTorch Foundation shares a keynote from PyTorch Conference North America featuring Cody Mazza-Anthony of Shopify. The talk walks through Shopify's continual learning loop with PyTorch and vLLM, showing how the company turns product expertise into robust evaluations, then mines, critiques, and repairs low-scoring conversations to train better models. Using Buyer Profile as a case study, the session demonstrates how small models can deliver meaningful results within a production feedback flywheel.
Together, these posts
- Open-Source AI Breakthroughs: Security, Agents, and TrustOpen-Source AI Takes Center Stage This week’s news digest reveals a clear trend: open-source AI is no longer a niche experiment—it’s becoming the backbone of enterprise innovation and security. From Mistral’s new open-weight model to Kubernetes powering AI agents in banks, … Read more
- Open Source AI: The Double-Edged Sword of AbundanceThe New Open Source AI Landscape Open source AI is no longer a niche experiment; it’s a defining force in how we build and deploy intelligent systems. A recent wave of developments suggests that the balance of power is shifting toward … Read more
- Open Source AI: Efficiency, Security, and CommunityWhy Open Source AI Is at a Tipping Point This week’s stories show open source AI moving from experimentation to production — and doing so on hardware that competes with the cloud. AMD’s work running a masked diffusion language model on … Read more
- Open Source News: Jira Alt, Ubuntu Rust, AI GovernanceOpenProject: A Rising Jira Alternative OpenProject is positioning itself as a compelling open-source alternative to Jira, with recent webinars highlighting its progress in agile project management, Jira migration tools, and resource capacity planning. The project’s roadmap includes step-by-step enhancements shaped by … Read more
- The Open Source AI Tipping Point: Why Closed Systems Are Losing GroundThe Quiet Revolution Something fundamental is shifting in the world of software. Across several recent examples, open source AI is no longer just a niche movement for hobbyists and academics—it’s becoming a serious alternative to proprietary, subscription-based tools. From enterprise orchestration … Read more
- The Open Source Lifecycle: From First Commit to Long-Term ImpactThe Journey of Open Source Projects Open source is more than code—it’s a living ecosystem where projects are born, evolve, and sometimes fade. But what determines whether a project thrives or dies? Drawing from a recent wave of developments, from security … Read more
- Open Source AI’s Double-Edged Sword: Security and SustainabilityThe Fragile Foundation of Open Source AI Open source AI is eating the world. From powering chatbots to automating enterprise workflows, open source models and tools are becoming the default for organizations that want flexibility, cost savings, and control. But as … Read more
- The Open Source AI Paradox: Control, Trust, and the Future of InnovationThe Open Source AI Paradox: Why Openness Is Both the Solution and the Problem Open source AI is often framed as a battle between open and closed systems, but a closer look reveals a deeper paradox: the same openness that drives … Read more