How ChatGPT Work agent lays foundation for OpenAI's AI super app ambition

OpenAI's ChatGPT Work combines coding, document creation and workflow automation in one interface, signalling a broader shift from chatbot to enterprise AI workspace

OpenAI has taken its biggest step yet towards turning ChatGPT into what could become an AI super app. Last week, the US-based artificial intelligence (AI) company unveiled ChatGPT Work, an AI agent that brings the capabilities of Codex across web, mobile and desktop.

 

According to the company, the new mode can gather context from the apps, files and workflows users choose, then turn that information into finished outputs such as documents, spreadsheets, presentations and web applications.

 

Alongside ChatGPT Work, OpenAI introduced GPT-5.6, its latest model family comprising Sol, Terra and Luna. The company said the models will roll out across ChatGPT, Codex and through its application programming interface (API).

 

 

Viewed individually, these are product updates. Together, however, they represent a broader strategic shift. Rather than positioning ChatGPT primarily as a chatbot, OpenAI is increasingly turning it into a unified workspace where users can research information, write code, generate documents and presentations, build websites and automate workflows without leaving the application.

The strategy mirrors the evolution of super apps in consumer technology, albeit with an enterprise focus. Instead of combining services such as payments, messaging and shopping, OpenAI is bringing together AI capabilities that were previously offered as separate products.

 

ChatGPT evolves from chatbot to AI workspace

 

For much of the past three years, ChatGPT has primarily functioned as a conversational interface. Users asked questions, generated text or images and, more recently, interacted with reasoning models.

 

ChatGPT Work changes that proposition by combining ChatGPT with OpenAI's AI coding tool, Codex, within a single interface.

 

OpenAI said the redesigned desktop application brings together Chat, Work and Codex. Chat is intended for everyday assistance, Work for complex multi-step tasks, and Codex for developers and technical professionals. Users can switch between these modes depending on the task, allowing the assistant to move beyond answering prompts to executing work autonomously.

 

Rather than interacting with separate AI products, users increasingly work through a single interface that selects the appropriate tools behind the scenes.

 

More importantly, ChatGPT Work is less about introducing another AI agent than redefining what ChatGPT itself becomes.

 

The company is steadily integrating conversational AI, coding, research, productivity tools, web publishing and autonomous agents into one platform. The redesigned desktop application, unified plugins directory, Scheduled Tasks, in-app browser, Computer Use capability and GPT-5.6 model family all reinforce that strategy.

The unified plugins directory connects ChatGPT with services such as Slack, Gmail, Google Drive, calendars and customer relationship management (CRM) platforms, while Scheduled Tasks automates recurring work. The desktop application also adds an in-app browser and Computer Use capability for interacting with websites and desktop applications.

 

OpenAI's AI super app strategy

 

Traditionally, a super app refers to a single application that combines multiple digital services.

 

India's Tata Neu, for example, brings together consumer businesses across the Tata Group, including BigBasket, Croma, Air India and Tata 1mg, into one platform. Users can shop for groceries, book travel, order medicines, access financial services and earn or redeem NeuCoins across the ecosystem.

 

Artificial intelligence is giving the concept a different meaning.

 

Rather than aggregating consumer services, AI companies are attempting to make a single interface the primary gateway for digital work. The objective is no longer simply answering questions but coordinating multiple AI capabilities from within one application.

 

Under this model, users no longer move between separate coding assistants, productivity software, search engines or website builders. Instead, an AI agent performs these tasks while interacting with external applications in the background.

 

ChatGPT Work represents OpenAI's clearest move in this direction.

However, OpenAI is not alone.

 

Earlier this year, Anthropic introduced Claude Cowork, an autonomous AI agent designed to plan and execute complex workplace tasks. Both products target knowledge workers rather than software developers, allowing non-technical users to benefit from increasingly capable AI coding systems.

 

Competition is therefore shifting away from building the best chatbot towards creating the most capable digital co-worker.

 

Major AI companies are increasingly developing agents that can:

• Plan projects independently.
• Execute multi-step workflows.
• Work across enterprise software.
• Write code when required.
• Produce complete business outputs rather than individual responses.

The commercial opportunity extends well beyond consumer subscriptions, with enterprise AI emerging as one of the industry's fastest-growing revenue streams.

 

Why GPT-5.6 matters

 

Alongside ChatGPT Work, OpenAI introduced GPT-5.6 in three variants:

 

Sol: Flagship model for advanced reasoning and coding.

Terra: Balanced model for general productivity.

Luna: Lower-cost model for high-volume workloads.

 

OpenAI said Sol establishes a new benchmark for intelligence and efficiency, delivering state-of-the-art performance across coding, knowledge work, cybersecurity and scientific applications while using fewer tokens and lowering costs.

 

The three-tier strategy reflects a broader shift in enterprise AI economics.

Rather than requiring customers to use the same frontier model for every workload, OpenAI is allowing organisations to match model capability with task complexity. Routine work can be handled by lower-cost models, while complex reasoning tasks can be assigned to Sol.

 

According to the company, GPT-5.6 is designed to deliver stronger performance per dollar while reducing inference costs, an increasingly important factor as enterprises expand AI deployment.

 

Enterprise AI is becoming a cost optimisation challenge

 

The announcement also highlights one of the AI industry's biggest challenges: economics.

 

Enterprise customers increasingly evaluate AI not only on model capability but also on the cost of deploying it across thousands or millions of requests. Token-based pricing can increase rapidly as usage scales, making cost optimisation a strategic priority.

 

OpenAI said GPT-5.6 has been designed to deliver faster performance at lower cost than competing frontier models while offering three model sizes suited to different workloads.

 

The strategy mirrors a broader industry trend towards reducing inference costs without sacrificing capability.

 

Lower pricing is becoming increasingly important as OpenAI competes not only with Anthropic but also with rapidly improving open-weight AI models that organisations can deploy within their own infrastructure.