OpenAI demonstrates voice steering, cloud handoffs, and personal agents for Codex at AI Engineer Code Summit
OpenAI outlined new interaction models for Codex, including voice control, cloud-based task persistence, and personal agents called 'dots' during a workshop at the AI Engineer Code Summit.
1 source · cross-referenced
- OpenAI showcased voice-based control for Codex, enabling hands-free task delegation and management.
- A revamped Codex Cloud version 2 allows tasks to persist in the cloud and be handed off from local execution.
- Personal agents called 'dots' were discussed as assistants for ongoing tasks and context management.
- Workshop presenters acknowledged usability challenges in cloud transitions and agent paradigms.
At the AI Engineer Code Summit, OpenAI engineers Dominik Kundel and Charlie Guo presented a workshop titled “New Ways of Working with Codex,” detailing recent updates to the coding agent. The session emphasized three interaction models: voice control, cloud-based task persistence, and personal agents referred to as 'dots.'
Guo and Kundel described a revamped version of Codex Cloud, released after DevDay in late September, which enables tasks to persist in the cloud and be handed off from local execution. Kundel noted that the transition to cloud-based workflows had previously been awkward for users, but the updated version aims to make cloud usage more viable by allowing users to publish an environment once and run multiple tasks in isolated instances. These environments can be shared within a company or kept private, and are configured by selecting accessible GitHub repositories.
Voice mode was presented as a hands-free interface for managing Codex tasks. Guo described using voice as a "control plane" to manage threads, check on work, pull data from other applications, and create new threads. Kundel shared that he uses voice dictation for iteration and to interact with his personal agent, noting that voice allows him to delegate work while commuting.
The workshop also touched on personal agents called 'dots,' which operate as persistent assistants across local, cloud, and development environments. Guo acknowledged that dots can feel unintuitive to developers but highlighted three use cases: as a safety net for missed items, as a "delegation sink" in Slack, and as a ubiquitous presence for ongoing context management.
- Nov 11, 2026 · The Verge — AI
OpenAI to launch verified AI Business Agents in ChatGPT for US e-commerce and service businesses
Trust79 - Nov 10, 2026 · Latent Space — swyx
NaiveAI releases Naive-N0.5-Flash coding model with ultrafast token throughput
Trust74 - Nov 10, 2026 · Latent Space — swyx
Resend CEO on why agents pick its email API over SendGrid and Cloudflare
Trust79