This weekly update covers exciting changes happening in computer coding with artificial intelligence agents. These are smart computer programs that help people write code without needing to do everything themselves. Two major AI tools recently became available in April: Claude Opus 4.6 and GPT-5.3-Codex, both claiming to be the best at helping people code. Claude Opus 4.6 is winning at finding and fixing mistakes in real programs from the website GitHub, while GPT-5.3-Codex is better at command-line work.

A big new tool called IBM Bob is changing how entire teams work on software. Instead of just helping people type code, IBM Bob acts like a real AI development partner that helps with every part of making software, from planning to testing. This is different from older tools because it helps with the whole job, not just writing code.

Computer experts are calling this new style "vibe coding," which means telling the AI what you want in regular English words, and the AI figures out the code. This makes programming much faster and easier, but it also brings new problems like security issues and messy code that's hard to fix later. Computer scientists say strong testing and human oversight are still super important.

These agentic AI tools are changing what jobs engineers need to do. There's now demand for new types of specialists like prompt engineers, people who are good at telling AI what to do. Companies are moving fast to use these tools because they help teams finish work quicker and with fewer mistakes.