Testing Out OpenAI's New "Super App"
The speaker demonstrates OpenAI's Codex as a potential 'super app' that consolidates multiple AI capabilities, showing how it can generate images, build websites, and follow complex multi-step instructions all within one interface.
Summary
The speaker discusses recent speculation that OpenAI is moving toward consolidating all their AI tools into a single 'super app' and argues that Codex appears to be positioned as this unified platform. They demonstrate Codex's expanded capabilities, which now include computer operation assistance, image generation, preference memory, and handling ongoing tasks. The demonstration begins with generating a mock-up image for a website selling surfboards and tacos, highlighting that this image generation now happens directly within Codex rather than requiring users to switch to ChatGPT. The speaker then shows Codex building a complete website based on the mock-up image, noting that it automatically applied front-end design skills and created SVGs instead of importing external images. The demonstration concludes with generating a specific background image featuring surfboards and a taco truck, with precise positioning instructions to ensure text readability - showcasing Codex's ability to understand and execute complex, contextual design requirements.
Key Insights
- The speaker believes Codex is positioned to become OpenAI's consolidated 'super app' that combines all their AI capabilities into one platform
- Codex now includes direct image generation capabilities, eliminating the need to switch between different OpenAI tools like ChatGPT
- The platform can automatically build complete websites from mock-up images using integrated front-end design skills
- Codex demonstrates advanced contextual understanding by creating SVGs instead of importing external images for web development
- The system can follow complex, multi-step design instructions while considering practical constraints like text readability and visual composition
Topics
Transcript
There's been a bit of talk recently that OpenAI is going to sort of consolidate everything into one single super app, like ChatGPT, their image generator, their code generator. And it really feels like that super app might end up being Codex. Codex can now operate your computer alongside you. It can now generate images. It can remember your preferences and it can take on ongoing and repeating tasks. Let's go ahead and say, generate a mock-up image of a website for a site that sells surfboards and tacos. The fact that it actually generated this image directly inside of Codex is new. You would have normally had to go to ChatGPT, generate it there, bring the image in. Build…
Full transcript available for MurmurCast members
Sign Up to AccessMore from Matt Wolfe
AI News: GPT-5.6 and the new Super App are a Massive Leap!
OpenAI released GPT-5.6, a major leap forward in AI capabilities, alongside the new unified ChatGPT work app integrating code, browsing, and agent features. The week also saw significant model releases from xAI (Grock 4.5), Meta (Llama Spark 1.1), and research from Anthropic on AI reasoning patterns, establishing a new competitive landscape in AI development.
I Built A Monetizable Business With AI
The creator built a monetizable finance dashboard in one day using Hyper Agent from Airtable, which deploys a team of AI agents that continuously monitor news, track funding rounds and IPOs, analyze market sentiment, and generate daily briefings without manual intervention. The system demonstrates how specialized AI agents can work together autonomously to create a functional business product.
The ONLY AI Benchmark You Need!
A developer created "Buccy Bench," a humorous yet functional AI benchmark that tasks different language models with drawing Gary Busey as SVG code rather than images. The benchmark tracks model evolution over time while measuring performance metrics like cost, tokens, and execution speed.
GLM-5.2 - The Open Model That's As Good As Opus!
A comprehensive review of GLM-5.2, an open-weight Chinese AI model with a 1 million token context window, demonstrating its capabilities for coding, document analysis, and agentic workflows at significantly lower costs than frontier models like Claude Opus and GPT-4.5. The speaker tests various use cases including website building, Chrome extensions, game development, and data organization, concluding it's valuable for long, code-heavy, token-expensive tasks despite not universally outperforming closed-source alternatives.
Don't Fall For This AI Trap
The speaker emphasizes that power users distinguish themselves by knowing what NOT to automate with AI, rather than automating everything. They argue that AI works best for clear, straightforward tasks but struggles with nuanced, artistic work requiring consistency—using their failed YouTube thumbnail automation as an example.