OpenAI and Anthropic rely on AI for coding, raising big job questions

AI is writing all our code

Boris Cherny, who leads Anthropic’s Claude Code effort, has said in public posts that he has been relying heavily on AI to generate code and that many of his recent pull requests were produced with model assistance.

Cherny has described shipping dozens of pull requests in a short period and said they were generated with model assistance, according to press coverage. The move has freed him from tedious coding, allowing him to focus on higher-level planning and creativity, signaling a shift in how software work is done.

Anthropic engineers embrace AI

Anthropic executives say that within engineering teams, many staff now use AI heavily to generate and edit code. At the World Economic Forum, Anthropic CEO Dario Amodei said he has engineers who report they no longer handwrite code, and he suggested models could do most engineering tasks within roughly six to 12 months.

This trend is part of a larger shift in the industry. AI tools like Claude Code are transforming coding from a painstaking process into a creative, prompt-driven workflow, making the old ways of programming feel outdated for many engineers.

OpenAI researchers follow suit

It’s not just Anthropic. OpenAI researchers report similar changes. On social media a researcher using the handle roon said they now rely on models to generate their code. That is a single user report and should not be taken as a statement of company policy or a measurement of overall developer practice

The adoption of AI coding tools is reshaping the software landscape. Developers are moving away from manual coding, using AI to handle detailed implementation, which could redefine career paths for programmers in the coming years.

AI is changing software fast

Experts say the software industry is evolving rapidly. Himanshu Tyagi of Sentient believes a large portion of code in the next decade will be AI-written. Claude Code is credited with accelerating this shift, making coding more accessible and efficient.

Even beyond coding, AI tools are being used for administrative tasks, project management, and internal communication. Engineers report higher satisfaction, since they can focus on creative work instead of repetitive coding chores.

Entry-level jobs face uncertainty

The rise of AI coding raises questions about the future of software roles, especially entry-level jobs. Traditionally, these positions taught new engineers the basics of coding. Now, AI handles much of the work, reducing hands-on learning opportunities.

Companies argue AI democratizes coding, letting even non-coders build products with natural language prompts. But the shift may require new approaches to training, mentorship, and skill development for the next generation of engineers.

Productivity soars with AI

Engineers like Cherny report dramatic productivity gains using AI. Routine coding is automated, freeing teams to innovate. Tasks that once took hours can now be completed in minutes, allowing engineers to focus on solving complex problems.

Even project management has been affected. AI systems can send reminders, update spreadsheets, and manage workflows. The combination of coding and admin automation is redefining what it means to be a software engineer today.

Industry adoption varies widely

Microsoft executives have said that roughly 20 to 30% of new code in some repositories is written with AI assistance, and Salesforce leaders have estimated that certain internal tasks are done with AI assistance in the tens of percent range.

Despite variation, leaders like Cherny expect AI-generated code to continue to grow. He predicts other companies will soon reach similar levels, and AI may extend into other computer tasks beyond programming.

Tools for coders and non-coders

Anthropic launched Cowork, a user-friendly agent built on the same agent architecture as Claude Code, and press coverage says the tool was developed quickly using the company’s agent tooling.

The tool’s success illustrates a new workflow in which AI acts as both an assistant and a builder. Teams can produce software faster while empowering non-technical staff to contribute, reshaping how projects are managed and executed.

AI coding still has limits

While AI handles most coding, it isn’t perfect. Models may make conceptual mistakes, over-complicate solutions, or leave unused code behind. Engineers need to review and refine outputs, ensuring quality and reliability.

Even with these limitations, confidence is growing. Experts expect AI-generated code quality to improve steadily, making human oversight more focused on design, strategy, and innovation rather than rote coding tasks.

AI changes hiring strategies

As AI handles more coding, companies like Anthropic are hiring differently. They now prefer generalists over specialists because AI manages many detailed tasks. Traditional programming skills are less critical, shifting the focus toward problem-solving and creative thinking.

This approach also impacts team structure. Engineers guide AI rather than write every line themselves, meaning mentorship and skill development now include how to work alongside AI effectively instead of only coding.

AI tools spark viral moments

Claude Code’s release created a viral reaction similar to ChatGPT’s debut. Engineers and non-coders alike shared excitement online, highlighting the AI’s ability to automate complex tasks while remaining accessible to users without deep programming skills.

The viral response reflects broader interest in AI productivity tools. It shows how quickly adoption can grow when the technology is both powerful and easy to use, creating momentum for industry-wide change.

AI supports admin and project tasks

AI isn’t just coding software. Engineers use Claude Code to handle project management, send Slack reminders, and update shared spreadsheets. These tools free engineers from repetitive tasks and improve team efficiency across the board.

By automating routine work, AI allows staff to spend more time on strategy, planning, and innovation. The result is a more creative and agile workflow that changes how teams operate daily.

Want to see how Claude is pushing the boundaries of creativity? Dive into the race with Gemini and Twitch in Pokémon Red.

AI reshapes software forever

AI is not just speeding up coding; it is redefining what it means to be a developer. Writing syntax is becoming less central, while guiding systems, reviewing output, and shaping ideas are taking priority. The job shifts from typing every line to steering the direction of the build.

That does not make human engineers less important. It changes where their value shows up. Creativity, system design, and judgment now matter more than memorizing code patterns, and those human strengths are what turn AI output into real, working products.

What happens when a startup fully commits to vibe coding? See how one Swedish AI company used it to smash every growth record.

What do you think about AI taking over coding? Share your thoughts.

This slideshow was made with AI assistance and human editing.

Don’t forget to follow us for more exclusive content on MSN.

Read More From This Brand:

• a16z finds startups spending on vibe coding
• The creator of Claude code breaks down where vibe coding falls short
• Using AI coding assistants may reduce speed for veteran developers