Picture this. You want to make a short AI playbook, with custom videos, a clean webpage, the works. Last year that meant ten browser tabs, three logins, and a lot of copy-pasting between tools that barely talked to each other.
Then I came across a post that flipped that whole mess on its head. The original poster showed off something that genuinely made me sit up: you can now generate images and videos straight from Claude. No tab-hopping. No exporting and re-uploading. Just chatting with Claude and watching real content come out the other side.
I was honestly surprised when I saw the videos in the post were made inside Claude itself, using connectors like the OpenArt MCP. Same idea for the rest of the build. The whole thing stayed in one place.
What changed: MCPs do the heavy lifting
Here’s the shift the creator points to. In 2025, building an AI playbook meant juggling 10+ separate tools. Now MCPs connect Claude directly to creative AI services. Once that link is set up, you just talk to Claude and create content inside that tool, without ever leaving your workflow.
The part that clicked for me is the context. When everything runs through Claude, the assistant already has your project files and your own Claude Skills loaded. So it knows what you’re working on. You also keep the freedom to decide how you want to use it. That combo is what makes it feel less like a toy and more like a real working setup.
A real example: playbook to webpage, no tool-switching
The author walked through turning part of an AI video playbook into a live webpage. The whole thing happened inside Claude Code, start to finish:
- Add the content to the project folder and launch Claude
- Generate the videos through the OpenArt MCP
- Build a webpage using those videos and the playbook
- Prepare deployment-ready files for the website
No jumping between apps. No babysitting five dashboards. Raw videos in, a deployed site out, all in one window.
Why this matters more than it sounds
The creator used to think node-based workflows were the ideal way to work with creative AI, especially for storytelling. After living with MCP-powered agentic workflows, the verdict flipped: there are far more practical, everyday use cases this way.
I like that honesty. Node-based canvases look impressive, but they can turn simple jobs into a wiring project. An agentic flow keeps you in conversation instead. You describe what you want, the assistant pulls the right tools, and the work moves forward.
How you can try this yourself
You don’t need a special setup to start. The whole point the original poster makes is that you can build one of these workflows on your own, using a daily AI assistant you already lean on.
- Pick one creative tool you use a lot, then check if it offers an MCP connector for Claude
- Drop your project files into a single folder so the assistant has full context from the jump
- Start small: generate one asset, like a short video or an image, before building anything bigger
- Chain the steps in plain language, from creating the content to prepping the files for publishing
The mind behind this post even notes a backup plan: Codex works as a Plan B if Claude isn’t your tool of choice. So the approach isn’t locked to one product, it’s a way of working.
What got me is how ordinary it makes something that felt advanced a year ago. One assistant, full context, real output. That’s a real change in how creative work gets done.
If you want the full walkthrough and the videos the author generated, check out the original LinkedIn post. It’s worth a look.