scripod.com
Inside OpenAI Enterprise: Forward Deployed Engineering, GPT-5, and More | BG2 Guest Interview

Overview

Shownote

Highlights

Transcript

Chapters

Pins

Inside OpenAI Enterprise: Forward Deployed Engineering, GPT-5, and More | BG2 Guest Interview

BG2Pod with Brad Gerstner and Bill Gurley

2025/09/11
BG2Pod with Brad Gerstner and Bill Gurley

BG2Pod with Brad Gerstner and Bill Gurley

2025/09/11
OverviewShownoteHighlightsTranscriptChaptersPins

Shownote

Open Source bi-weekly convo w/ Bill Gurley and Brad Gerstner on all things tech, markets, investing & capitalism. This week, guest host Altimeter’s Apoorv Agrawal explores how OpenAI is reshaping enterprise with Sherwin Wu, Head of Engineering OpenAI Platf...

Highlights

This episode of BG2, hosted by Apoorv Agrawal with OpenAI’s Sherwin Wu and Olivier Godement, delves into the expanding role of AI in enterprise environments. The conversation moves beyond consumer-facing chatbots to examine how organizations are leveraging advanced AI systems for mission-critical applications across industries—from telecommunications and drug discovery to national security.
04:30
B2B is core to OpenAI's mission as it can distribute AGI benefits, enabling enterprises to make real-world impacts.
10:46
Working with T-Mobile has helped improve OpenAI's models.
11:30
Faster drug development could save millions of lives
13:47
GPT-3 was physically installed on an air-gapped supercomputer at Los Alamos due to security constraints.
17:01
95% of AI deployments fail, requiring top-down buy-in and bottom-up execution.
25:23
Autonomy has three parts: perception, reasoning, and scaffolding
28:32
GPT-5 significantly reduces hallucinations while excelling in coding and reasoning tasks
30:02
GPT-5 is excellent at instruction following but can be too literal, requiring clearer prompts.
33:00
Real-time API now integrates speech input, intelligence, and text-to-speech seamlessly
35:31
The model can intelligently handle voice input and output in real time.
40:20
RFT combines reinforcement learning with fine-tuning to create best-in-class, use-case-specific models.
1:00:54
OpenAI's sprint to launch GPT-5 showcased its strengths in science, infrastructure, and talent.
1:07:50
GPT-4’s reasoning ability made previously impossible queries solvable.

Chapters

Intro
00:00
OpenAI’s Enterprise Mission: Beyond ChatGPT
01:50
Case Study: T-Mobile <> Voice & Support
06:00
Case Study: Amgen <> Accelerating Drug Development
11:30
Case Study: Los Alamos National Lab
13:45
Why 95% of AI Deployments Fail?
17:00
Physical vs Digital Autonomy: Scaffolding & Infrastructure
20:30
GPT-5: Release, Benchmarks vs Behavior
26:00
GPT-5 Feedback: Instruction Following, Hallucinations, Code Quality
30:00
Multimodality: Text, Voice, and Video
33:00
Audio: Realtime API vs Stitched Audio
35:30
Model Customization & Reinforcement Fine-Tuning (RFT)
38:00
Rapid Fire: Long/Short Picks
43:00
Highlights and Lowlights @ OpenAI
1:03:00

Transcript

Sherwin Wu: We literally had to bring the weights of the model physically into their supercomputer. Apoorv Agrawal: In San Francisco, you could take a car from one part of SF to the other fully autonomously. As opposed to the digital world, I can't book a...