Docker, Inc

Shipping beyond localhost? You can now use Kanvas - a new extension for Docker Desktop, to translate Compose to Kubernetes - without drowning in YAML. Use Designer mode to import Compose, add Ingress and Secrets, validate, and reuse templates. Switch to Operator mode to inspect Pods and Services, stream logs, open terminals, generate traffic, and overlay Prometheus metrics. Plus: A built-in policy engine turns AI-assisted designs into safe, deployable blueprints.

If you're looking to upskill your engineering teams, take a page from Rabobank. They just launched their internal Docker Championship Program - a hands-on initiative focused on developer enablement and container security. 50+ engineers joined the kickoff event to learn best practices for building secure, optimized images. Led by Docker, this kind of program is a strong model for teams investing in modern software workflows. We’re excited to keep collaborating - and we’re looking forward to seeing how this inspires other teams to kick off their own internal programs.

🔥 Missed re:Invent? One of the biggest highlights was the launch of Kiro - AWS’s new agentic IDE for AI-powered coding. Agents are becoming more powerful, but they can still do unexpected things, like deleting everything, leaking credentials, or executing code you didn’t ask for. This hands-on walkthrough shows how to give Kiro (and other agents) real guardrails using Docker’s MCP Catalog and Sandboxes - so you can keep them productive and contained. Build faster, stay sandboxed, and keep your dev loop safe - even in YOLO mode. #Docker #AWSreInvent #Kiro #MCP #DevTools #AI #AgenticAI #LLM #SecureAI

Most AI discussions obsess over new model releases. Ep 9 of AI Guide to the Galaxy digs into what actually breaks when developers try to use them. Daniel Han from Unsloth AI shares: - Why many high profile models ship with broken chat templates, wrong tokens, or architecture mistakes, and how Unsloth ends up fixing them - How their automatic compiler generates optimized kernels for any new model, removing per-model engineering - Practical guidance for developers on fine tuning, reinforcement learning, data needs, and the simple hardware rule of thumb: if you can run it, you can usually tune it If you’re building on open source LLMs, this episode exposes the real-world issues and workflows devs should know. ▶️ Watch AI Guide to the Galaxy Ep 9 now: https://lnkd.in/eMT-GgV7 #AIGuideToTheGalaxy #AI #ML #DevTools

San Francisco, look up. Docker + E2B just hit the skyline. Smarter, safer AI starts here with Docker’s MCP tools running natively in your E2B sandboxes. 🎥 Drone footage of the billboard 👇 #Docker #E2B #AgenticAI #DevTools #AI #SanFrancisco #CloudNative

ChatGPT is great at “say,” not “do.” This guide shows you how to change that safely. Using the Docker MCP Toolkit, you’ll learn to give ChatGPT secure, containerized capabilities: it can scrape websites, persist and analyze data, run code, and even open pull requests. No duct tape, no guesswork, just Docker guardrails for real-world AI workflows. In this new guide, you’ll learn how to: - Connect ChatGPT to a containerized MCP Gateway - Add plug-and-play MCP servers like SQLite, GitHub, Firecrawl, and Stripe - Build a full Competitive Repricing Agent that scrapes prices, analyzes them, and updates Stripe prices automatically. Why read it? Most GPT workflows break at the edge of reality - fragile, insecure, and impossible to scale. With Docker MCP, you move from clever prompts to real execution: isolated, reproducible, and credential-aware. Durable AI, built the right way.

The Docker MCP Gateway just got a deep dive courtesy of AI Labs. In a recent video, AI Labs breaks down how Docker’s MCP Gateway helps AI agents run smarter, not harder. If you’re experimenting with tool chaining, agent workflows, or LLM orchestration, this is worth your time. Key takeaways: - Dynamic tool discovery and selection - “Code Mode” for secure, efficient automation - Streamlined agent performance with fewer tokens and less bloat Watch the full breakdown: https://lnkd.in/einEX6H3 #Docker #AI #MCP #DevTools #AgentOps #LLMs #DeveloperExperience #DockerDesktop #OpenSource

Modular monoliths. Real-time CVE patching. Safer AI agent workflows. The December issue of Docker Navigator is out - packed with tools, insights, and real-world stories to help you build smarter, more secure software. From AI agents to supply chain security, here’s what teams are doing with Docker right now. #Docker #DevTools #Security #AI #Newsletter #SupplyChain #Containers #LLM #AgenticAI

AI agent development is still too complex. That’s why we’re changing it. We want you to have the tools to build great agents without fighting your stack. Ones like Docker Model Runner, cagent, 300+ MCP tools like MCP catalog and Toolkit, Agentic Compose, and sandboxes are all redefining how AI gets built. And we have the numbers to prove it. According to theCUBE Research, with Docker: - 87% of developers cut AI project setup time - 90% reduced failures - 97% accelerated time-to-market for new products. #Docker #AI #AgenticAI #DevOps #theCUBEResearch

Most teams still think fine tuning is “ML wizard” territory. 🪄 Daniel Han from Unsloth AI thinks that fear is holding developers back. In AI Guide to the Galaxy Ep 9, we talk about: - How local models have rapidly closed the gap with frontier systems across real workloads - How Unsloth gets 2–3x faster training and major memory savings using mathematical optimizations instead of hardware hacks - Why dynamic quantization retains more model quality compared to static approaches - The practical differences between fine tuning and reinforcement learning, and why even a small number of examples can be enough to see improvements If fine tuning has felt too difficult, this episode breaks down the actual constraints around model size, data, and hardware in plain engineering terms. ▶️ Watch AI Guide to the Galaxy Ep 9 with Daniel Han now: https://lnkd.in/eC5wDD35 #AIGuideToTheGalaxy #ML #AI #DevTools