OpenAI has introduced new features aimed at helping people learn and teach with ChatGPT Work and Codex, according to a company announcement. The update is positioned as making it easier for organizations to build internal fluency with AI tools directly inside the products, rather than relying entirely on external training resources or documentation.
Details on the specific mechanics — whether this includes guided tutorials, in-product coaching, structured lesson formats, or admin-facing tools for teaching teams — were not fully specified in the source material and should be treated as unconfirmed until OpenAI publishes fuller documentation or a product changelog. What is confirmed is that the update touches both ChatGPT Work, OpenAI's enterprise-oriented product tier, and Codex, its coding-focused AI system, suggesting the learning features are meant to span both general knowledge-work use cases and more technical, developer-facing ones.
This kind of update fits a broader pattern from OpenAI and competing AI vendors: as adoption of generative AI tools inside companies matures past the experimentation phase, vendors are increasingly focused on the "last mile" problem — not whether the model is capable, but whether the humans using it know how to get consistent value from it. Training and onboarding friction has been a recurring theme in enterprise AI rollouts. Surveys of enterprise AI adoption have repeatedly found that a meaningful share of licensed seats go underused, often because employees were given access to a tool without a clear sense of what to do with it in their specific role.
For a small or mid-sized B2B company, this changes little about strategy but potentially something about execution cost. Historically, getting a support team or a sales team fluent in an AI tool required either a paid training vendor, an internal champion who self-taught and then trained others, or simply hoping people figured it out. If ChatGPT Work now includes built-in teaching material or guided learning paths, that reduces the marginal cost of the first step: getting a team from zero to competent baseline usage.
It's worth being clear-eyed about what this doesn't solve. Learning to use ChatGPT well is a different problem from designing a workflow where AI reliably handles a specific business process — triaging inbound support tickets, drafting first-pass sales replies, summarizing calls into CRM notes. The former is about individual skill; the latter is about system design, error handling, and integration with existing tools. Companies that conflate the two often end up with employees who are personally more productive with AI, but no organizational change in how work actually flows through the business.
The announcement did not include pricing changes, availability timelines by region, or details on whether these learning features are included in existing ChatGPT Work subscriptions or offered as an add-on. Companies currently evaluating or already using ChatGPT Work should check OpenAI's official documentation directly for specifics before making rollout decisions based on this announcement alone.