Urgent.News

What's breaking now, across thousands of outlets.

AI

How My OpenAI Dot Starts Real Claude Code Sessions on My PC

Orion is my OpenAI Dot: an always-on agent I message from ChatGPT. It lives in the Codex desktop app on my Windows PC, which lets it run commands there. Claude Code is where my engineering runs. I wanted Orion to be able to hand Claude a job without me at the keyboard. On 2026-10-07 it did. I asked Orion to have Claude use one of my agent phones to Google a question and bring back what Google…

On October 7, 2026, Orion, an always-on agent known as my OpenAI Dot, successfully initiated a Claude Code session on my Windows PC. This agent runs within the Codex desktop app, enabling it to execute commands on the machine. The goal was to allow Orion to delegate tasks to Claude Code without me needing to manually interact with it.

I instructed Orion to have Claude use one of my phone numbers to conduct a Google search for the term "what is OpenAI dot" and subsequently relay the AI Overview text back to our chat. The entire process was documented through a series of timestamps, providing insight into the sequence of events.

Key moments in the operation included a health check at 2:05:54 PM Pacific Time (PT), Orion's prompt being written to a file at 2:06:52 PM, and the service accepting the start request at 2:06:58 PM, all occurring within a matter of seconds. The service then facilitated the leasing of an Android phone, which executed the Google search at 2:07:04 PM.

After six minutes, the session concluded, with Orion providing a detailed report on its findings. Throughout this process, stringent safety measures were implemented to prevent any unauthorized actions such as purchases, account changes, or modifications to unrelated work. The demonstration was carried out entirely on a single machine, with various checks and balances in place to ensure the integrity and security of the operation.

Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at dev.to →

More in AI

Aventra Group secures ISO 42001 artificial intelligence certification

Technology firm Aventra Group achieved international ISO 42001 certification across its infrastructure to unify AI governance models with existing information security management frameworks

  • Company integrates ISO 27001 and SOC 2 Type II frameworks into unified governance baseline.
  • Aventra Group's certification aims to build trust and enable global AI scaling.

More from Thursday 8 October →