GPT-6 Astra and the "AGI Era": A Marketer's Reality Check

Every few years a lab says the word. On 3 September OpenAI released GPT-6 Astra, and its president Greg Brockman told reporters it is "not unreasonable to feel that we are now in the AGI era." I am a researcher and a founder, so I hold two thoughts at once: this is a genuinely capable system, and "AGI" is a claim, not a measurement.

What happened

According to Fox Business, OpenAI released Astra first to enterprise customers with early access, with ChatGPT Plus, Pro, Business and Enterprise users, the API and Amazon Web Services to follow within days. OpenAI says the model is state of the art on computer use, browser use, software engineering, cybersecurity and science, and that it can carry out routine computer tasks such as filling in online forms, updating customer records, organising calendars and doing research, as well as creating documents, spreadsheets and presentations.

Sam Altman called it the company's "most aligned model ever". OpenAI also said Astra represents a significant jump in cyber capabilities and meets the "Critical" threshold under its Preparedness Framework. The article quotes OpenAI's own definition of AGI: "highly autonomous systems that outperform humans at most economically valuable work." It cites no specific benchmark scores.

Why it matters

Strip away the label and the practical shift is clear. The capabilities OpenAI highlights are not about writing better sentences. They are about doing work in software: filling forms, updating records, moving through browsers and spreadsheets. That is the daily texture of marketing operations. Campaign setup, reporting, CRM hygiene, competitor monitoring, feed management. A model that is good at computer use attacks the hours, not just the words.

That is also why the AGI framing deserves caution. OpenAI's definition is about outperforming humans at "most economically valuable work". Whether that is true for your team is an empirical question you can test, and the answer will differ by task. A model can be excellent at assembling a performance report and still weak at deciding which creative idea will move a sceptical audience in Jakarta.

What my research says

I have written in Forbes on whether machines will replace human creativity and on closing the gap between marketing and AI systems. My consistent position is that the gap is rarely raw intelligence. It is grounding: does the system know your audience, your data and your constraints, and can it show its reasoning?

That is what our research targets. In MindFuse (ACM Multimedia 2025) we built explainable generative AI that combines creative click-through-rate scoring with LLMs to co-create marketing strategy from real ad data. In Fusing Predictive and Large Language Models for Actionable Recommendations in Creative Marketing (ACM TOIS 2025) we paired performance prediction with LLMs so that the recommendation comes with a reason. The reason is the point. A more powerful general model makes the language better and the actions faster. It does not by itself tell you why one ad will beat another for a specific audience. That still needs measurement against real outcomes.

So my reality check is not "Astra is overhyped." It is "Astra is a stronger engine; your evidence and your explanations are still your job."

What marketing leaders should do now

• Pick three time-consuming operational workflows, such as weekly reporting, campaign setup or CRM updates, and pilot them with a computer-use model under supervision.
• Measure against a baseline. Time saved, error rate, and how often a human had to step in. "Felt faster" is not a result.
• Separate execution from judgement. Let agents execute well-defined steps; keep creative and strategic decisions tied to data you can explain.
• Mind the permissions. A model that can update customer records can also update the wrong ones. Give agents the narrowest access that works and log every action.
• Ignore the label in procurement. Ask vendors for task-level evidence on work like yours, not for adjectives.

Whether or not this is the "AGI era", it is clearly the era in which AI does operations. Teams that build careful, measured workflows around that will get the gains. Teams that buy a word will get a bill.

FAQ