OpenAI DevDay!: demoing GPT-4 Turbo, "GPT Store" potential, and more with Sunny Madra | E1841

Episode Summary

Executive Summary: The hosts review OpenAI DevDay, focusing on customizable GPTs, a planned GPT store, lower API costs, faster responses, and a larger context window. They see GPTs as a potential new app platform that could change how people use the internet, while warning that OpenAI’s copyright legal-cost pledge may invite lawsuits. Demos of AI products highlight opportunities in white-label chatbots, email assistance, browser research, and natural-language automation, alongside rising competition from Google and xAI.

Main Topics: OpenAI GPTs and a prospective app marketplace: OpenAI’s new GPT-building tools let users configure assistants with instructions, selected models, and uploaded documents, without requiring coding. The hosts see a planned store and possible usage-based creator payments as a major evolution of plugins and a route to a new app ecosystem. Copyright Shield and legal exposure: OpenAI said it would cover certain legal costs for ChatGPT Enterprise and API customers facing copyright claims. The hosts questioned how the protection would work and warned it could attract lawsuits, while noting the possibility that OpenAI is confident in its training-data practices. GPT-4 Turbo performance and cost changes: The new GPT-4 Turbo preview was described as faster and cheaper, with a larger context window and more recent training knowledge. The hosts saw these improvements as important for production applications, while testing raised concerns that long transcripts were not yet fully analyzed. Startup opportunities beyond OpenAI’s platform: Demos of Droxy AI and Mindy illustrated opportunities to build white-label assistants and email-based AI workflows. The hosts argued that startups can differentiate through control of branding, domain, distribution, monetization, and integrations rather than relying entirely on OpenAI’s ecosystem. AI in browsers and business automation: Brave’s page summarization and Zapier’s natural-language workflow setup were presented as examples of bringing AI into existing tools. The hosts envisioned browser histories becoming searchable knowledge bases and automation becoming accessible to less technical users. Competition, adoption, and the changing internet: The hosts suggested that GPT apps could divert some activity from websites and search, while broadening access to useful software creation. They also discussed xAI’s Grok and urged Google leaders to demonstrate faster progress across products such as Gmail, YouTube, and Chrome.

Key Arguments: GPTs could make specialized assistants as easy to create and distribute as apps, unlocking experimentation by non-developers and potentially changing how users access information and services. An app-store-style GPT marketplace could create new opportunities for creators, but its compensation model was not yet defined; the hosts preferred a transparent, use-based payment formula over an unspecified revenue pool. OpenAI’s promise to cover legal expenses could reassure enterprise customers, but unclear eligibility may encourage opportunistic copyright lawsuits. The hosts emphasized that using material from a third-party crawl does not automatically remove responsibility for its underlying rights. Faster, cheaper GPT-4 access and a larger context window can improve application economics, though early testing showed the new model did not reliably summarize an entire long podcast transcript. AI startups can survive platform competition by offering capabilities platforms may not prioritize, including white-label deployment, customer-owned domains, flexible monetization, and integration into existing workflows. Embedding AI in email, browsers, and automation tools can reduce friction and make complex tasks accessible to nontechnical users; the strongest products should learn users’ preferences and fit their current workflows. Competition from xAI and other developers may accelerate the whole sector, while established companies such as Google risk falling behind if their product teams do not ship and demonstrate AI features quickly.

Data Points: Time since ChatGPT launch: A little over 11 months - The hosts framed OpenAI’s first developer conference as arriving less than a year after ChatGPT launched. OpenAI DevDay livestream audience: About 40,000–50,000 viewers - The hosts used the figure to emphasize how early the developer ecosystem remains compared with large established platforms. GPT-4 Turbo context window: 128K tokens - Presented as the model’s larger capacity for taking in documents and other material; the hosts loosely compared tokens to words. Blended GPT-4 Turbo cost reduction: About 2.75 times cheaper - The show’s notes characterized the blended cost relative to the previous GPT-4 model. GPT-4 Turbo API pricing: $0.01 per 1,000 input tokens and $0.03 per 1,000 output tokens - Pricing figures cited by the hosts for the new model. GPT-4 speed comparison: About as fast as GPT-3.5 Turbo in the demonstrated response - The demo was used to illustrate the improved responsiveness of GPT-4 Turbo. LinkedIn membership: 1 billion users - Mentioned in an advertisement, not as a central topic of the discussion. Grok development timeline: Less than 90 days - The hosts cited this as an impressive development pace, while tentatively judging the product to feel several months behind competitors.

Pivotal Quotes: "This is a very dangerous thing to do. It's literally like waving a red flag in front of bulls." — Host: A warning that OpenAI’s legal-cost pledge could attract drive-by attorneys and copyright lawsuits. "Instead of consuming the internet through the end sites, you'll find the bot to interact with." — Sundeep Madra: Describing how GPT apps might change the way users discover and consume online services. "Now, everybody's going to be good at everything." — Host: Summarizing the broader potential of AI tools to make skills such as writing, automation, and basic design accessible to more people.

Implications: GPT marketplaces and embedded assistants could shift discovery and transactions away from websites, while creating new creator and startup opportunities. Success will depend on reliable performance, clear rights and payment rules, and products that offer ownership or workflow advantages beyond major platforms.