OpenAI explodes to Codex! It's available to non-coders, and 900 million people are happy to have “AI employees”.”

Last night.OpenAILive Stream AnnouncementAI programmingintelligent bodyThree Codex updates, respectively, are smart body plug-ins, fixed-point modifications, and document one-click generation of interactive sites.

OpenAI also revealed that Codex's weekly active users have reached5 million, an eight-fold increase from the beginning of the year.

Among them.Intelligent Body Plugins (Agent Plugins)Packaged for the enterprise six exclusive job skills package, enterprises do not need to program access plug-ins can let the “senior AI employees” work;AnnotationsThis feature helps users to make precise changes to the generated content, “changing it wherever they want”;SitesFunctions allow users to transform any document into a functional preview of interactive websites, applications and share it with colleagues and clients.

These features will gradually go live on Codex today, with site functionality going live in a preview version for enterprise and commercial teams today, and gradually opening up to other subscription plan users in the near future.

OpenAI programIntroducing Codex intoChatGPT, available to all subscribed users. In February, OpenAI announced that ChatGPT has more than 900 million weekly active users.

In addition, OpenAI announced itsAll Frontier models and Codex now support one-click deployment on the AWS platform, which offers organizations two ways to use it: either by calling OpenAI models directly on Amazon Bedrock, or by bringing OpenAI's software-engineered intelligence, Codex, to AWS via Codex on Amazon Bedrock.

At the end of the livestream, OpenAI co-founder and CEO Sam Altman showed up and said that this update to OpenAI Codex is all about helping organizations integrate all their products and services while reducing costs. He predicted that the next focus of the AI industry will be toActive AI that operates autonomously around the clock, which is his preferred direction for the most worthwhile advance layout in the coming year.

Altman mentioned that today OpenAI is moving into theThird stage of developmentAnd at this stage they are aiming toMaking AI accessible to all.

▲Sam Altman, Founder and CEO of OpenAI, and Denise Dresser, Chief Revenue Officer of OpenAI (from left to right)

I. Codex trifecta, debuts 6 character-specific plugins

Denise Dresser, OpenAI's chief revenue officer, revealed that the enterprise business currently accounts for $40% of OpenAI's revenue and will account for $50% by the end of the year. this part of the business is a key focus for OpenAI.

To improve the efficiency of users' smart body collaboration, it has updated three features for Codex.

First up are the Intelligent Body Plugins.OpenAI debuted six character-specific plugins that will gradually go live as codecs and will soon be available in ChatGPT.

Each plugin bundles apps, skills, instructions, and workflows for the relevant role, with a total of 62 popular apps and 110 skills.

The Data Analytics plugin helps analysts and business teams use data to answer questions and create reports and dashboards using tools like Snowflake, Databricks Genie, Hex, and Tableau.

▲ Utilize data analytics plug-ins for business performance analysis

The Creative Production plugin helps marketing and creative teams turn briefs into review-ready material, and teams can also create campaign boards, create and optimize display ads, and more using tools like Figma, Canva, Shutterstock, Picsart, and Fal.

Sales plug-ins help sales teams incorporate customer contextual information into their workflow for moving deals forward, with tools such as Salesforce, HubSpot, Slack, Outreach, Clay, Rox, and Actively available to find high-priority customers and signals, prepare for customer meetings, complete follow-ups, and more.

▲Use the sales plugin to generate sales channels and prepare for meetings

There is also a product design plugin, a public stock investment plugin and an investment banking plugin.

Enterprises can adapt plug-ins to their own workflows or build and share custom plug-ins for their own systems.OpenAI will soon launch more role-specific plug-ins for corporate finance, private equity, marketing strategy, strategy consulting, and legal, and in the future it will also support enterprises to create and deploy their own plug-ins directly in Codex and ChatGPT.

Next are the annotations. For Codex outputs, the user can select any of them and have them interpreted and edited.

Based on the annotation feature, users can point out parts that need improvement and tell Codex what needs to be changed. For example, checking the navigation bar in the website and asking Codex to update the fonts; marking up a chart on a slide and asking for clearer labels to be added, and so on.

▲Optimizing tables with the annotation feature

Finally, there are Sites. With Sites, users can turn any content created by Codex into something safe and easy to share. For example, financial forecasts can be turned into real-time business dashboards for the entire company, and product plans can be turned into working prototypes for teams to review together.

▲ Translate financial forecasts into real-time business dashboards for the entire company

According to Alexander Sergian, Head of Enterprise Business at OpenAI, all of this improvement is due to the performance advancement of OpenAI models, while the tokens used are decreasing in parallel. Users require only 1/3 the number of tokens to get the same quality output with Codex and GPT-5.5.

Second, OpenAI enters the third stage of development, to let everyone use AI

OpenAI Chief Revenue Officer Denise Dresser also spoke with OpenAI Co-Founder and CEO Sam Altman at the end of the broadcast.

Altman mentioned that he sensed in his interactions with customers that the biggest need for organizations was to integrate all of their products and services to form a unified set of workflows within the organization. Because these features were previously siloed from each other, users were often unsure of when to use chat, codecs, or hidden APIs, and were unable to integrate all of the context and required plug-ins.

Another important thing is cost. According to Altman, there's even a stigma forming in AI circles now that a lot of companies spend their full-year 2026 budgets in one quarter. In the face of this problem, OpenAI is investing in the development of more models that allow people to get more value from less investment.

The third thing that this extends to is that many users don't think they're fully tapping into the value of AI, and many don't really know how to ask the right questions, which can also lead to high token consumption. So what OpenAI is doing is automating everything, without requiring the user to understand for themselves what the AI can do for me and how it works.

Regarding the relationship between OpenAI and the enterprise, Altman argues that OpenAI sees itself as a new kind of infrastructure, much like cloud services used to be.

The focus of OpenAI's R&D this year is broadly divided into two categories, one for conversational large models and one for intelligent body products that can autonomously complete complex tasks.

In Altman's view, the next focus of the AI industry is on active AI that operates autonomously 24/7, which is his preferred direction for the most worthwhile advance in the coming year.

Speaking about OpenAI's journey, he revealed that OpenAI is entering its third phase, after two phases of transitioning from a research lab to a product company, the landmark of which was the introduction of ChatGPT three and a half years ago.

Six and a half years ago, OpenAI's first big token user used about 100,000 tokens per month, and now this usage has reached the level of global per capita usage; the top users consume about 100 million tokens per month, and the overall token usage has skyrocketed by a factor of 1 million in six and a half years.

If the current growth trend continues and is extrapolated back another six and a half years, Altman predicts that the above token usage will become the global benchmark for per capita token consumption or per capita AI usage. At that point, everyone will be more aware of the infrastructure pressures, which is exactly the third stage of development OpenAI is in.

OpenAI's goal is to make AI accessible to all, so that they can experience both top performance and affordable functionality, while guaranteeing an adequate supply of arithmetic power.

Conclusion: Non-developers flood inAI Programming Tools, users are growing 3 times faster than programmers

Today, more and more non-developers are using AI programming tools.According to OpenAI, while Codex started as a software development tool, today non-developers such as analysts, marketers, and operations personnel account for about 20% of Codex users, and are growing at more than three times the rate of developers.

Such a user structure also makes Codex and other AI programming tools toward more low-code, automation progression, its product development focus shifted to adapt to the enterprise business people with natural language quickly landing needs.

Article source: Wisdom