Article · 2026-08-03 · 6 moments

🎙️ How I AI: ChatGPT Codex Voice + browser + Sites: an expert’s AI workflow | Nick Baumann (OpenAI)

Your weekly listens from How I AI, part of the Lenny's Podcast Network ✦ AI generated

Mechanism◆

ChatGPT Codex Voice doesn't just answer questions — it can actually operate your computer while you keep talking, running several tasks in parallel via separate threads.

Nick Baumann demos the new Codex Voice interface live: he triggers it with a hotkey, the orb reads his screen through Appshots, and Codex spins up separate threads to book a flight, file an expense report, and check the calendar at the same time.

transcript

Claire: ChatGPT Codex Voice doesn't just answer questions. It can actually operate your computer while you keep talking. Nick Baumann demos the new interface live: he triggers it with a hotkey, the orb reads what's on his screen through Appshots, and Codex spins up separate threads to book a flight, file an expense report, and check his calendar at the same time.

Mechanism

ChatGPT Sites looks closer to a real deployment platform than a lightweight demo, with SQL database, file storage, environment variables, and email-based access controls — genuinely useful for internal tools, small apps, and private resources.

Claire and Nick build and publish a live How I AI tips site, organized by tool and job function, in roughly the time a CMS template setup would take — showing Sites is a real deployment platform, not just a demo.

transcript

Claire: ChatGPT Sites looks much closer to a real deployment platform than a lightweight demo. It includes a SQL database, file storage, environment variables, and email-based access controls. During the episode, Claire and Nick build and publish a live How I AI tips site, organized by tool and job function, in roughly the time it would normally take to set up a CMS template. For internal tools, small apps, and private resources, that starts to feel genuinely useful.

provides context · 1

Example

Heartbeats in ChatGPT Work offer a practical glimpse of always-on agents: Nick gets notified about suspicious charges, nearby packages, and emails needing attention — all set up without writing code.

Nick demonstrates always-on agents in ChatGPT Work, where heartbeats proactively surface notifications for suspicious charges, package arrivals, and email attention — set up entirely without code.

transcript

Claire: Heartbeats in ChatGPT Work offer a practical glimpse of always-on agents. Nick gets notified when a suspicious charge appears, a package is nearby, or an email needs his attention—and he set it all up without writing code.

provides context · 1

Example

Nick's video-editing workflow is one of the clearest examples of AI cutting real production work: he records dozens of raw clips, talks through the rough story, and a custom plugin transcribes, scans, selects the strongest takes, and assembles finished vertical videos overnight.

Nick records 50 or 60 raw clips, uploads them, talks through the story he wants, and a custom plugin handles transcription, shot selection, and assembly overnight — handling what would normally eat up most of a content team's day.

transcript

Claire: Nick's video-editing workflow is one of the clearest examples of AI cutting real production work. He records 50 or 60 raw clips, uploads them, talks through the rough story he wants, and lets a custom plugin transcribe everything, scan the footage, choose the strongest takes, and assemble finished vertical videos overnight. What would normally eat up most of a content team's day is largely handled while he sleeps.

explains mechanism · 1

Mechanism

Thread-forking is one of Codex's most powerful features: it can open a new thread, branch from an existing one, or pull context from previous conversations on its own, without the user manually organizing everything first.

When a task gets complicated, Codex forks threads on its own — branching from existing work and drawing context from prior conversations, so users don't have to organize everything manually. Nick frames this as one of Codex's most powerful yet least understood features.

transcript

Claire: One of Codex's most powerful features is also one of its least understood: thread-forking. When a task gets complicated, Codex can open a new thread, branch from an existing one, or pull context from previous conversations on its own. Users don't have to manually organize everything first.

extends · 1provides context · 1

Claim◆

Voice may be the best way to give AI context, because talking freely removes the pressure to be precise — the 'yapper's API' — and people who talk to AI give it more context, which leads to better work.

Claire dubs voice the "yapper's API": people often get more useful results when they talk freely than when they stare at an empty prompt box trying to be precise. The pattern holds across two-person voice chat, mobile dictation, and Codex's orb interface.

transcript

Claire: Voice may be the best way to give AI context because it removes the pressure of figuring out exactly what to type. Claire calls this the "yapper's API," and the idea is simple: people often get more useful results when they talk freely than when they stare at an empty prompt box trying to be precise. Across two-person voice chat, mobile dictation, and Codex's orb interface, the same pattern keeps showing up—people who talk to AI tend to give it more context, and more context usually leads to better work.

gives example · 1provides context · 1supports · 1

Highlight slides

Voice as the 'Yapper's API'✦ from: Voice may be the best way to give AI context, because talking freely removes the pressure to be precise — the 'yapper's API' — and people who talk to AI give it more context, which leads to better work.ChatGPT Codex Voice operates your computer✦ from: ChatGPT Codex Voice doesn't just answer questions — it can actually operate your computer while you keep talking, running several tasks in parallel via separate threads.The pattern holds across interfaces✦ from: Voice may be the best way to give AI context, because talking freely removes the pressure to be precise — the 'yapper's API' — and people who talk to AI give it more context, which leads to better work.

Related episodes

Article · ByteByteGo Newsletter

AI Customer Support at Scale: The Travel Industry’s $Billion Bet

→

Video · The Cognitive Revolution

Intelligence on the Edge: Liquid AI's Ramin Hasani on the Search for Device-Native Foundation Models

→

Article · The Pragmatic Engineer

What is “loop engineering?”

→