Three new free Academy courses teach workers a tidy three step model: prompt, then workflow, then agent. The curriculum itself is the product.
OpenAI's newest product isn't a model. It's the training that teaches a workforce how to use OpenAI.
On June 12, 2026, the company announced three new OpenAI Academy courses: AI Foundations, Applied AI Foundations, and Agents and Workflows, framed as preparation for what the company calls "the next era of work." The curriculum walks learners from understanding AI, to applying it to recurring tasks, then to directing structured agent workflows. It is co-developed with BCG, Accenture, and BBVA, the kinds of names that sign off on enterprise rollouts.
That last detail is the story. The model vendor is now writing the corporate syllabus for its own products, and the consultancies that historically trained on rival stacks are co-signing that curriculum. Three new courses sound like an education announcement. In practice, OpenAI is inserting itself into how organizations define AI-augmented work, which is a different kind of competitive moat than model capability alone.
The three-track structure maps to a clear progression. AI Foundations establishes the vocabulary. Applied AI Foundations pushes learners toward using AI on recurring, repeatable tasks. Agents and Workflows goes further, teaching workers to orchestrate structured agent workflows rather than just run them. Underneath that progression sits an implicit three-step model: prompt, then repeatable workflow, then agent-assisted workflow. It is a tidy way to organize skill-building, and it is also a tidy way to organize product adoption. Each step requires fluency with OpenAI's tools.
That is the bargain on offer. Free training from the vendor whose products an organization is being asked to adopt, with top consultancies lending credibility to a curriculum that doubles as a deployment playbook. For workers, the appeal is real: structured learning paths that connect to concrete tasks, with names like BCG and Accenture on the syllabus. For organizations, it lowers the friction of standing up internal training. For OpenAI, the curriculum becomes a funnel. Workers who learn to think in agents on OpenAI's terms are workers who reach for OpenAI when the next project lands.
The partner list deserves attention. BCG and Accenture are not bystanders. They are co-authors of the curriculum, and they have their own stakes in the framing. Consulting firms benefit when the model of "AI-augmented work" settles around a vendor-aligned curriculum, because their engagements tend to grow with the surface area of the rollout. BBVA, the Spanish bank, signals that the program is meant to extend beyond the United States. The mix is the message: a global consulting and finance chorus, all pointing in roughly the same direction.
None of this means the courses are bad. Foundations-level material on prompting and applied use has value regardless of the source. The Agents and Workflows track, in particular, is likely to be useful for teams trying to standardize how they ship automations. The risk is not that the content is wrong. The risk is that workers and organizations absorb a vendor's view of how work should evolve as if it were neutral education.
The framing of "the next era of work" is worth pausing on. It is OpenAI's own positioning, and the announcement treats it as a settled destination. Adoption is not inevitable. The implicit three-step model, from prompt to repeatable workflow to agent-assisted workflow, is a useful organizing idea, but it is one possible organizing idea. Workers and organizations can adopt the skills, adapt the framing to their own tools and constraints, or refuse the premise that vendor-authored training is the right place to learn how their jobs should change.
What to watch next: how the Agents and Workflows track aligns with OpenAI's current Agents SDK and product state, since the term "agent" has been a moving target. Whether OpenAI publishes enrollment, completion, or outcome data, or whether the program stays a press-release story. And whether competing enterprise AI training offerings, from Anthropic, Google Cloud Skills Boost, Coursera enterprise, and Microsoft Learn, start to publish their own curriculum philosophies, so workers and procurement teams can compare what each vendor wants the future of work to look like.