
Tech • IA • Crypto
OpenAI is overhauling ChatGPT into a full AI “super app” centered on task automation and coding agents to drive enterprise revenue and compete with Anthropic.
OpenAI is preparing a major redesign of ChatGPT aimed at transforming it from a question-answer tool into a full-service AI agent. The system is expected to handle tasks across work and daily life, including scheduling, communication, and automation. The long-term vision is for users to assign goals while the AI selects tools and completes tasks autonomously.
The company’s coding tool, Codex, is becoming central to this strategy. It will be embedded directly into ChatGPT, enabling software control, workflow automation, and general computer operations. OpenAI is expanding Codex beyond developers with role-specific plugins for areas such as sales, creative production, and investing.
Early implementations focus on everyday productivity. Users could connect email or Slack accounts to draft and send messages, or integrate calendars to summarize meetings and prepare daily agendas. Automated workflows, such as morning task briefings, illustrate the shift from answering questions to executing actions.
OpenAI expects transitional interfaces with buttons and tool selection to eventually disappear. Future versions aim to interpret intent automatically and decide whether tasks run locally, in the cloud, or through external integrations, reducing user involvement in tool selection.
Advances in GPT‑5.5 have improved multi-step reasoning and task execution, reducing the need for manual guidance. This allows Codex to operate more reliably as an autonomous agent, increasing trust among developers and enterprises.
Codex adoption has surged, with its user base reportedly growing sixfold within two months of a desktop app launch. Weekly active users exceeded 5 million, while usage has been increasing by roughly 5% per day. Enterprise revenue tied to Codex has also shown sharp gains.
Despite having close to 1 billion users, most ChatGPT usage remains free, creating revenue pressure. Around 2 million businesses now use OpenAI products, contributing roughly 40% of revenue, with expectations to reach 50%. Coding and productivity tools are seen as easier to monetize than casual chat features.
OpenAI has deprioritized some consumer features, including pausing shopping tools and reportedly discontinuing certain media-generation efforts, to focus on enterprise services and agents. Internal restructuring has merged ChatGPT, Codex, and API teams under a unified leadership.
Rival Anthropic has gained traction with developer-focused tools like Claude Code, pushing OpenAI to accelerate its efforts. Industry benchmarks previously suggested Anthropic had an edge in programming, raising stakes in a category closely tied to AI advancement.
Anthropic’s annualized revenue has reportedly reached $47 billion, while OpenAI’s exceeds $30 billion. Both companies are increasingly focused on enterprise adoption and monetization, signaling convergence toward investor-driven growth strategies.
The rivalry extends to talent and hardware. A key contributor to OpenAI’s in-house chip project recently joined Anthropic, highlighting competition in AI infrastructure. OpenAI is collaborating with Broadcom on a large-scale accelerator system expected to roll out from 2026 through 2029.
As AI agents gain more autonomy, OpenAI has introduced a restricted lockdown mode for sensitive environments. It disables features like browsing and file downloads to mitigate risks such as prompt injection attacks that could expose data or trigger unintended actions.
OpenAI’s overhaul of ChatGPT reflects a broader shift toward AI agents that perform real work, with Codex at the core of monetization and enterprise strategy amid intensifying competition with Anthropic.