Hey Devs!
A busy week for builders: Gemini 4 Argon opens to developers, voice agents gain faster decisions, and GitHub expands security scanning and repair. Here are updates to catch up on.
Sponsored by call4me
Give your AI assistant a phone. Ask your AI assistant to move a dentist appointment or get a quote, and let it make the call. call4me connects Claude, ChatGPT, Codex and Cursor to real business phone lines through MCP, then returns the result and transcript to your conversation. See call4me in action.
Gemini 4 Argon opens to developers
Google made Gemini 4 Argon available through the paid Gemini API, with access in Antigravity for Google AI Ultra subscribers. It targets complex software engineering and long workflows such as code migrations, making it a candidate to test against your hardest repository tasks.
Read more: https://blog.google/innovation-and-ai/technology/developers-tools/build-with-gemini-4-argon/
GPT-Live connects voice to Decisions
OpenAI added native GPT-Live integration with its public-beta Decisions API, allowing decisions as input or output transcripts stream. Live can call Decisions directly, removing an application round trip; your code still executes the actions, useful for responsive voice controls and interactive avatars.
Read more: https://developers.openai.com/api/docs/guides/decisions-voice
GitHub ships AI Scan and agentic autofix
AI Scan for pull requests is now GA, extending security coverage to PHP, Bash, Terraform, and other areas CodeQL misses; findings remain advisory and consume AI Credits. Separately, agentic autofix can repair default-branch alerts, rerun CodeQL, and open a draft PR for review, with eligible Enterprise Cloud security licensing required. That gives teams broader detection and a patch checked against CodeQL before human review.
Which update would you try first in a real project?
Cheers, proflead! ;)


