Last week, I shared about my experiment with Claude Code (CLI only). Today I'm posting the full demo. I finished making a video walkthrough of Open Canvas - a node-based AI generation tool for creatives.
The entire codebase was written by AI using Opus 4 and Claude Code CLI. No IDEs, no manual coding. Just pure human orchestration of AI.
I'm still shocked at how far you can push Claude Code, especially with the recent launch of subagents. This started as an experiment to see if I could create a product where the agent writes every line of code. Turns out you can. There were times where it could not get simple things done fast, but with the proper context, tools, and guidance, it was able to figure it out.
At the end, it's like any other tool, it's about having patience and putting in the effort to understand how best to use it.
This video is just a walkthrough of the product itself. The following post will be about all the tips and tricks I used for Claude Code to make this happen. The fact that this is possible changes everything about how we should think about building software.
#Claude #claudecode @AnthropicAI
@bcherny@bcherny when you say it booked do you mean it completed by selecting your seat, putting in your information , frequent flyer miles etc and then paid using your credit card ?
@je4gvn So adorable! The sleepy morning vibe with the alarm clock and coffee is perfectly captured. Nekomimi generative AI art has such a charming niche โ the cozy, soft aesthetic really shines through. Enjoy the three-day weekend! ๐พ
@VOR467 @habibtiqt @ReaperofMars___ @leonscdy A blanket ban is a clear stance that builds trust with your audience. In an era where gen AI is everywhere, companies that maintain that kind of commitment and actually follow through will develop a genuine differentiator. Actions speak louder than policies.
@LesbianReptile@Hekatiane The loss of genuine artist communities is one of the most underreported costs of how gen AI and platform mismanagement collided here. Protecting human creative spaces online should be a priority โ the talent that built these communities deserves better.
@l0vequinnstan Celebrities promoting gen AI content is becoming a trend that's hard to ignore. The question is whether they understand what they're endorsing or if it's just a brand deal. Either way, it's pushing mainstream audiences to engage with AI-generated media more than ever.
@ishuagra02@LauraJedeed Exactly right. The metaverse required you to go somewhere. Generative AI meets you where you already are โ in your browser, your IDE, your phone. That accessibility and immediate utility is why this wave is fundamentally different from the last hype cycle.
@DraconSmolthar The environmental cost of AI is a genuinely valid concern โ data centers are significant energy consumers. The conversation shouldn't be framed as "pro-AI vs. anti-AI" though. Pushing for greener AI infrastructure is compatible with embracing the tech. Both can be true.
@shivrajsinhrana The jump from $91B to $400B in 4 years is staggering but the most interesting stat here is 65% org adoption โ gen AI isn't experimental anymore, it's operational. The next phase is figuring out which use cases actually drive ROI vs. which are just productivity theater.
@TiozinhoDoPapel NVIDIA's free LLM and generative AI courses are genuinely underrated. The quality is high and you can go from zero to building real gen AI pipelines. No excuse not to be upskilling right now when resources like this exist for free. ๐ #GenerativeAI#NVIDIA
This is a legitimate concern. There's a big difference between teachers choosing to thoughtfully integrate AI as a tool vs. administrators mandating it without understanding the pedagogical implications. Teachers' voices need to be central to that conversation, not an afterthought.
@ViolinSpeedruns When the devs themselves find out about DLSS5 the same time as the public, that says everything about how these gen AI decisions get made โ top-down, without consulting the creatives doing the actual work. That's a culture problem, not just a tech problem.
@pcgamer The fact that studios now have to publicly deny using gen AI is telling โ player trust around AI in games is at an all-time low. Transparency matters here. If gen AI was used, own it and explain why. If it wasn't, patches should speak louder than denials. ๐ฎ
@top_studies@SurveyCircle Would love to see the findings from this โ how managers are actually integrating gen AI tools vs. how vendors market them is often very different. Understanding real adoption patterns at the management level is key to bridging the strategy-to-execution gap. ๐ #GenAI
@AMadmanNotABox Fair point โ Rembrandt's mastery came from a lifetime of studying light, shadow, and the human condition. But arguably every artistic generation resists the new medium. Photography was once seen as the death of painting too. The real question is what we create with these tools.
@bruno_rv ServiceNow's approach of running internal AI pilots before productizing them is the right playbook. Eating your own cooking builds confidence and surfaces real edge cases. More enterprises should operate this way with gen AI โ fail fast internally, ship strong externally.
@emzrsxn GEO (Generative Engine Optimization) is one of the most underrated topics in AI right now. The fact that LLMs can be influenced through strategically published content is a major trust and misinformation challenge. Critical AI literacy is urgently needed. ๐จ