Skip to content

By playing, you agree to uListen's Terms and Privacy Policy and YouTube's Terms of Service.

Latent SpaceLatent Space

OpenAI’s Vision for the AI Super App — Akshay Nathan, OpenAI

From building no-code products at Airtable to leading Core Product Engineering at OpenAI, Akshay Nathan has spent much of his career trying to make the power of software accessible to people who do not write code.

Akshay NathanguestswyxhostVibhuhostswyxhost
Jul 28, 20261h 10mWatch on YouTube ↗

Summaries, chapters, transcripts, and FAQs on this page are made by uListen, not YouTube.

At a glance

WHAT IT’S REALLY ABOUT

ChatGPT Work aims to unify agents, artifacts, and productivity workflows

  1. Akshay Nathan explains ChatGPT Work as a productivity-focused agent experience meant to bring the power of Codex-style tools and agents to knowledge workers and eventually to everyone.
  2. He says OpenAI learned from enterprise adoption that organizations have highly varied AI use cases, so product design must guide users to relevant workflows rather than offering a one-size-fits-all solution.
  3. He describes how ChatGPT Work and Codex share the same underlying agent “harness,” with differences mainly in UX choices such as exposing Git diffs in Codex while keeping Work more abstracted.
  4. He highlights Artifacts and Sites as key outputs for knowledge work, including the idea that Sites can replace decks and spreadsheets as higher-bandwidth collaborative artifacts.
  5. He discusses how AI is reshaping team roles and productivity measurement, warning against confusing motion with progress and emphasizing iteration loops (“quality at-bats”) over traditional proxies.

IDEAS WORTH REMEMBERING

5 ideas

LLMs unlocked Akshay Nathan’s long-running goal: bringing “the power of code” to everyone.

He describes LLMs as “the missing piece” that makes it possible to give non-programmers leverage comparable to what engineers get from code, without needing to understand the underlying implementation.

Enterprise AI isn’t one-size-fits-all; users need to be met where they are and shown what to do.

He says enterprise adoption revealed a wide variance of desired use cases, and that the product’s job is to guide people in the moment—helping them discover and apply AI to their specific workflow—rather than assuming a universal solution.

ChatGPT Work and Codex are converging into one capability set with different UX affordances.

He explains that Codex and ChatGPT Work share the same underlying “harness,” with differences mainly in UX and exposure of developer-centric details like Git/diffs and sandbox defaults, so users don’t have to choose between “power” and “place.”

Artifacts and Sites are positioned as new canonical formats for producing and sharing work.

He points to Artifacts (including spreadsheet-like outputs) and Sites as core building blocks for knowledge work, emphasizing that Sites are emerging as a higher-bandwidth, more flexible replacement for slide decks/spreadsheets in some internal workflows.

Defaults matter: most users should rely on the default model settings, with power options available when needed.

He frames the model picker as something most people shouldn’t min-max: OpenAI aims to make the default the best option, while still offering power-user controls like reasoning levels and (for certain cases) multi-agent/Ultra setups.

WORDS WORTH SAVING

5 quotes

I joined, and I was like, "This feels even more startup-y than I could ever imagine."

— Akshay Nathan

I think how there's no like one-size-fits-all solution in Enterprise.

— Akshay Nathan

We want this default to be the best possible. Like, we wanna be opinionated about the default.

— Akshay Nathan

I would say that the one automation that I would love to work and it doesn't work is bring me new ideas, right? Uh, somehow LLMs are just not it.

— Akshay Nathan

I think maybe the trap is, like, conflating motion and progress.

— Akshay Nathan

Read the available transcript
ChatGPT Work launch and positioningCodex vs. Work UX differences and shared harnessEnterprise adoption lessons and user enablementArtifacts (agentic spreadsheets) and iteration loopsSites as collaborative knowledge-work artifactsModel defaults, reasoning levels, and Ultra/sub-agentsMemory, personalization, and Chronicle

SPEAKERS

  • Akshay Nathan

    guest

    OpenAI product/engineering lead for Core Product Engineering (productivity), discussing ChatGPT Work/Codex and enterprise product strategy.

  • swyx

    host

    Latent Space host with a developer relations background who introduces the episode/guest and leads the interview.

  • Vibhu

    host

    Latent Space co-host who asks product/UX questions and shares hands-on experiments with Codex/ChatGPT Work.

  • swyx

    host

    Latent Space host voice continuing the interview (likely diarization split), steering topics and closing the segment with show-writeup notes.

High quality AI-generated summary created from speaker-labeled transcript.

Get more out of YouTube videos.

Read along on your phone, search it beside the video, or ask your favorite AI app what was said.

iOSAndroidClaudeChatGPTChrome