ChatGPT Is Getting Its Biggest Upgrade Ever

OpenAI is aggressively repositioning ChatGPT from a chatbot into an AI super app and agentic platform, with Codex as the centerpiece for making coding, automation, and computer tasks accessible to everyone. The company is seeing explosive traction—five million weekly active users and strong enterprise revenue—and has pivoted away from browser-imitation agents toward code-based execution as the more effective interface for real-world tasks, a shift driven largely by Anthropic’s Claude Code leapfrogging OpenAI in programming ability. In response, OpenAI restructured teams, reallocated resources from projects like Sora to coding and enterprise tools, and introduced features like a desktop app and lockdown mode to lower adoption barriers and address security risks. Meanwhile, the rivalry with Anthropic has intensified into a direct revenue race, highlighted by a high-profile chip-team engineer leaving OpenAI for Anthropic and the two labs competing fiercely over developers, enterprise customers, and the agent layer. Ultimately, OpenAI’s broader strategy is to evolve beyond the chatbot era, convert free users into paying customers through Codex, and secure its position in the race toward superintelligence.

• ▶ 0:07 OpenAI’s goal is to turn ChatGPT into the main AI layer for your digital life, evolving it from a chatbot into an AI super app.
• ▶ 1:34 Codex is the most important piece: OpenAI plans to integrate it deeply into ChatGPT so coding, workflow automation, and computer tasks become available to everyone, not just developers.
• ▶ 3:25 The core shift is from “answer my question” to “handle this task,” with agents eventually deciding which tools to use automatically—powered by GPT-5.5’s improved multi-step abilities.
• ▶ 4:24 Claude is gaining major attention because Anthropic has been releasing useful new features like Claude Code, Artifacts, Skills, and Connectors, making it one of the most practical AI tools for turning ideas into real work.
• ▶ 4:53 A common frustration is that many people say “learn Claude” without offering a clear, practical path to actually use it effectively.
• ▶ 5:00 The sponsor’s “Claude-a-thon” solves this with a two-day live workshop covering deep research, dashboards, presentations, connectors, custom agents, and more, with 1,000 free seats available.
• ▶ 5:48 Developers are beginning to trust Codex for end-to-end coding tasks, with examples like Notion engineers finding Codex silently fixing a long-standing bug, making it feel like an integrated workflow tool rather than a novelty.
• ▶ 6:31 OpenAI lowered the adoption barrier with a Codex desktop app, offering a familiar GUI while still supporting power users and mobile check-ins, helping normalize background-agent behavior.
• ▶ 7:12 Codex shows explosive growth—record search interest, sixfold user growth since the desktop app launch, 5 million weekly active users, and enterprise Codex revenue up 50% week over week—positioning it as a key monetization lever for OpenAI’s enterprise push.
• ▶ 8:46 OpenAI pivoted away from browser-imitation agents like Operator, concluding that writing code is a more effective interface for AI to complete real tasks — reframing Codex as the foundation for a general AI agent.
• ▶ 9:30 The strategic shift was driven by Anthropic’s Claude Code preview in February 2025, which overtook OpenAI in programming ability — a critical category because better coding accelerates AI research and the race toward superintelligence.
• ▶ 10:30 OpenAI restructured by creating a dedicated Codex team, merging ChatGPT, Codex, and API teams, and reallocating resources away from projects like Sora toward enterprise tools, coding agents, and revenue-generating services.
• [11:22-11:54] Anthropic and OpenAI are in a direct revenue race, with Anthropic reportedly reaching $47B annualized revenue in May and OpenAI passing $30B—signaling fierce competition over revenue, developers, enterprise adoption, and the agent layer.
• [11:54-12:45] Clive Chan, OpenAI chip project employee #002, left for Anthropic, calling its talent, values, and ambition impressive and saying "It's time to build"—another high-profile talent shift between the rival labs.
• [12:45-13:10] The talent move highlights OpenAI's chip collaboration with Broadcom: OpenAI designs chips and systems, Broadcom handles accelerator and networking work, with first racks expected in H2 2026 and the full plan running through 2029.
• ▶ 14:15 OpenAI introduces lockdown mode to reduce prompt injection risks for users handling sensitive data, restricting web browsing, image display, deep research, agent functions, and file downloads.
• ▶ 15:13 The mode is intentionally not for everyone—it targets organizations needing stricter protection against data exfiltration via prompt injection.
• ▶ 15:24 The broader strategy has three goals: convert free users to paying customers via Codex, race Anthropic on coding agents/enterprise/AI chips/talent, and evolve ChatGPT beyond the chatbot era to justify paid adoption.

• ▶ 0:02 ChatGPT’s Super App Overhaul and Codex Integration (0:02 - 4:21) - OpenAI plans a major ChatGPT transformation, shifting from chat to personal agent with Codex plugins and future model upgrades.
• ▶ 4:21 Claude’s Rising Popularity and Live Workshop (4:21 - 5:48) - Claude attracts attention, and the sponsor segment promotes a Claude-a-thon live workshop.
• ▶ 5:48 Codex Adoption and OpenAI’s Enterprise Push (5:48 - 8:32) - Developers trust Codex, the desktop app lowers barriers, and OpenAI focuses on growth and monetization.
• ▶ 8:32 Agent Competition and Strategic Restructuring (8:32 - 11:24) - Operator shifts to code as the interface; Anthropic rivalry and revenue pressure drive OpenAI restructuring.
• ▶ 11:24 Revenue Race, Chip Development, and Talent Shift (11:24 - 14:15) - Anthropic’s revenue climbs, OpenAI co-develops chips with Broadcom, and a notable researcher leaves for Anthropic.
• ▶ 14:15 Lockdown Mode and Final Wrap-Up (14:15 - 16:11) - ChatGPT’s lockdown mode and overhaul strategy are reviewed, followed by a channel promotion.