I Replaced Claude Design… 100% UNLIMITED
Jack Roberts introduces Open Design, an open-source, locally-run alternative to Claude Artifacts with an Apache 2.0 license, allowing users to use any AI model they choose. The tool replicates Claude's design capabilities while removing usage limits, supporting 71+ brand design systems, and enabling multi-model workflows. Roberts demonstrates building a SaaS analytics dashboard and integrating ChatGPT image generation within the local environment.
Summary
Jack Roberts presents Open Design, a trending GitHub repository he positions as a free, unlimited replacement for Claude Artifacts (which he refers to as 'Claude Design' throughout). He opens by criticizing Claude Artifacts' limitations: rapid context exhaustion, roughly one meaningful project per week, outputs that look like 'AI slop,' being locked to Opus 4.7 only, and being confined to the Anthropic ecosystem.
Open Design differentiates itself on several fronts. It carries an Apache 2.0 license, making it commercially safe for agency and freelance use — unlike other clone repos Roberts says have appeared on YouTube but restrict commercial usage. It runs entirely locally, meaning nothing is uploaded to external servers. It includes over 71 brand design systems (including Apple, Nike, PlayStation, Coinbase, and others), five export formats (HTML, PowerPoint, PDF, ZIP, standalone HTML), and direct deployment to Vercel.
Roberts walks through a live demonstration using the tool inside an environment he calls 'anti-gravity,' cloning the GitHub repo and launching the locally hosted UI. He configures a project called 'Vantage,' a SaaS analytics dashboard, selecting dark theme, stripe-style density, and interactive animations. The one-shot output includes live data visualizations with MRR charts, workspace retention, and a world heat map. He then switches mid-project from Claude Code to OpenAI Codex to demonstrate seamless model swapping.
He also shows how to integrate OpenAI's image generation API to add a Ghibli-style illustration into the dashboard, highlighting that multi-model collaboration is a core strength of Open Design versus Claude Artifacts' single-model constraint. Roberts compares the models to Marvel's Avengers, arguing each has unique strengths and combining them produces better results than relying on one.
His recommended workflow is to use Anthropic's Opus model for initial design architecture, then switch to cheaper or free models (like DeepSeek V3) for iterative refinements and scaling. He notes the repo is only 12 days old and may have minor bugs, but found none personally. He concludes by recommending Open Design for agencies, freelancers, and multi-model operators, while suggesting Claude Artifacts remains fine for beginners or those comfortable with its limitations.
Key Insights
- Roberts claims Claude Artifacts exhausts its context so quickly that users can realistically complete only one meaningful design project per week, even on the $200/month plan, which he says provided no more usage than the $20/month plan.
- Roberts argues Open Design's Apache 2.0 license is a critical differentiator from other Claude clone repos appearing on YouTube, because those alternatives prohibit commercial use, making them effectively useless for client work or product sales.
- Roberts demonstrates that Open Design allows mid-project model switching — he switches from Claude Code to OpenAI Codex without losing design context — and argues this is impossible within the Claude Artifacts ecosystem.
- Roberts recommends a specific two-phase workflow: use Anthropic's Opus model to establish the initial design architecture and system, then hand off iterative refinements and scaling to cheaper models like DeepSeek V3, which he claims delivers about 95% of the performance at roughly 1/100th the cost.
- Roberts highlights that Open Design supports direct Vercel deployment from within the app UI, meaning a locally generated design can be published as a fully functioning web app by simply entering a Vercel token — bypassing any external design-to-code pipeline.
Topics
Transcript
[0:00] Claude Design is amazing, but the limits are brutal and it can stop you dead in your tracks. Open Design is a clone of Claude Design, and it is 100% local. And in this video, I'll show you exactly how to use this to build any website, app, or design that you want with any model you want to, even 100% for free, so you can stop wasting time and start growing your business. And if you don't know who I am, my name is Jack Roberts. I bought them for my last tech startup with thousands and thousands of customers. [0:30] Now I build my own uh companies and show you the exact things that work. So if…
Full transcript available for MurmurCast members
Sign Up to AccessMore from Jack Roberts
100 hours of Hermes Agent lessons in 23 minutes
Jack walks through advanced features of Hermes Agent, an AI personal assistant, covering memory systems, background tasks, scheduled cron jobs, model switching, and integration with tools like Obsidian, GitHub, and various AI models. The video aims to help users unlock capabilities beyond basic chatbot usage. Key themes include connecting external memory systems, delegating tasks to specialized AI models, and building a persistent, context-aware personal assistant.
Claude Code = $10,000 Beautiful Slides
Jack demonstrates a GitHub-based system using Claude Code to generate professional slide decks from any website URL by codifying 20 universal design principles. The system uses Firecrawl to extract brand DNA from websites and can integrate AI-generated images via APIs like Krea.ai. The result is polished, brand-consistent HTML presentations created in minutes with minimal input.
The most valuable thing I got from working at McDonald's
The speaker reflects on their first job at McDonald's at age 17, earning £5 an hour. The most valuable takeaway was the realization that they never wanted to work such a job again, reinforced by calculating how long it would take to earn a million pounds at that wage.
Claude Code Memory System = CHEAT CODE
Jack Roberts presents a three-tier Claude memory system designed to give AI tools persistent context across all platforms and sessions. The system consists of short-term identity memory, mid-term project memory via structured folders and claude.md files, and long-term memory using either Obsidian or Pinecone for archiving conversations and expert knowledge. The goal is to eliminate the 'amnesia' problem where AI loses context between chats.
ChatGPT ads aren't actually a bad thing. Here's when I want them.
The speaker argues that ads in ChatGPT are not inherently bad, particularly when users are actively seeking product recommendations. They distinguish this as 'pull advertising,' where the user initiates the request and ChatGPT synthesizes results based on provided context.