Inside OpenAI Enterprise: Forward Deployed Engineering, GPT-5, and More | BG2 Guest Interview
Episode
68 min
Read time
2 min
Topics
Productivity, Health & Wellness, Investing
AI-Generated Summary
Key Takeaways
- ✓Forward Deployed Engineering: OpenAI embeds engineers directly with enterprise customers like T-Mobile to build custom integrations, connect models to internal CRM systems without APIs, design evaluation frameworks, and optimize latency for production voice calls—moving beyond simple API access to full system implementation.
- ✓Reinforcement Fine-Tuning Advantage: RFT allows enterprises to create best-in-class models for specific domains using their own data and gradable tasks. Rogo achieved superior results on financial document parsing, while Accordance reached state-of-the-art performance on CPA-level tax tasks using OpenAI's RFT product.
- ✓GPT-5 Design Philosophy: Development prioritized customer feedback over benchmark saturation, focusing on instruction following precision, reduced hallucinations approaching zero, code quality improvements, and behavior tuning. Months of embedded customer work shaped the model's tone, style, and practical business application capabilities beyond raw intelligence.
- ✓Real-Time Voice API Architecture: Speech-to-speech models eliminate the three-step pipeline of speech-to-text, reasoning, and text-to-speech. This preserves emotional signals, accents, and tone while reducing latency and interruptions. T-Mobile deployed this for automated customer support calls requiring natural human-sounding interactions at scale.
- ✓Healthcare AI Acceleration: Pharmaceutical companies like Amgen represent the highest-potential AI transformation sector due to massive structured data volumes, document-heavy regulatory processes, and technical R&D culture. Automating drug approval documentation and research analysis could double medication development rates, impacting hundreds of millions of lives.
What It Covers
OpenAI's enterprise platform leaders discuss forward deployed engineering, GPT-5 development, real-time voice API, reinforcement fine-tuning, and major deployments at T-Mobile, Amgen, and Los Alamos National Labs transforming customer support, drug development, and national security research.
Key Questions Answered
- •Forward Deployed Engineering: OpenAI embeds engineers directly with enterprise customers like T-Mobile to build custom integrations, connect models to internal CRM systems without APIs, design evaluation frameworks, and optimize latency for production voice calls—moving beyond simple API access to full system implementation.
- •Reinforcement Fine-Tuning Advantage: RFT allows enterprises to create best-in-class models for specific domains using their own data and gradable tasks. Rogo achieved superior results on financial document parsing, while Accordance reached state-of-the-art performance on CPA-level tax tasks using OpenAI's RFT product.
- •GPT-5 Design Philosophy: Development prioritized customer feedback over benchmark saturation, focusing on instruction following precision, reduced hallucinations approaching zero, code quality improvements, and behavior tuning. Months of embedded customer work shaped the model's tone, style, and practical business application capabilities beyond raw intelligence.
- •Real-Time Voice API Architecture: Speech-to-speech models eliminate the three-step pipeline of speech-to-text, reasoning, and text-to-speech. This preserves emotional signals, accents, and tone while reducing latency and interruptions. T-Mobile deployed this for automated customer support calls requiring natural human-sounding interactions at scale.
- •Healthcare AI Acceleration: Pharmaceutical companies like Amgen represent the highest-potential AI transformation sector due to massive structured data volumes, document-heavy regulatory processes, and technical R&D culture. Automating drug approval documentation and research analysis could double medication development rates, impacting hundreds of millions of lives.
Notable Moment
Physical autonomy through self-driving cars has surpassed digital autonomy despite higher safety requirements because autonomous vehicles benefit from standardized infrastructure like roads and traffic laws, while AI agents operate without scaffolding in unstructured digital environments with constantly evolving interfaces and requirements.
Episode Transcript
Literally had to bring the weights of the the model, physically into their supercomputer. 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 ticket online right now. Physical autonomy is ahead of digital autonomy in 2025. I think AI agents are, like, reeling day one here. Like, Chargebee T only came out in 2022. The slope, I think, is incredibly steep. I actually do think self driving cars have a good amount of scaffolding in the world. You have roads. Roads exist. They're pretty standardized. You have stoplights. AI agents are just kind of dropped in the middle of nowhere. We'll start with long short game. I'm short on, the entire category of, like, tooling, Evals products. Healthcare is probably the industry that will benefit the most from AI. I think I'm AJ Pelt. You're definitely AJ Pelt. The first one was the realization in 2023 that I would never need to code manually like ever ever again. Hey, folks. I'm Apoorv Aggarwal. And today at the OpenAI office, we had a wide ranging conversation about OpenAI's work in enterprise. I have with me the head of engineering and head of product of the OpenAI platform, Sherwin Wu and Olivia Goodman. OpenAI is well known as the creator of SharedGPT, which is a product that billions across the world have come to love and enjoy. But today, we dive into the other side of the business, which is OpenAI's work in enterprise. We go deep into their work with specific customers and how OpenAI is transforming large and important industries like health care, telecommunications, and national security research. We also talk about Shervin and Olivia's outlook on the next what's next in AI, what's next in technology, and their picks both on the long and short side. This is a lot of fun to do. I hope you really enjoy it. Well, two world class builders, two people who make building easy. Sherwin, my Palantir 2013 classmate, tennis buddy, with two stops at Quora and Opendoor through the IPO before joining OpenAI. Before ChatGPT, you've now been here for three years and lead engineering for all OpenAI platform. Olivier, former entrepreneur, winner of the Golden Lama at Stripe, where you were for just under a decade, and now lead all of the product at, OpenAir platform. That's right. Thanks for doing it. Thank you. Thanks for having us. You know, as a shareholder, as a thought partner, kicking ideas back and forth, I always learn a lot from you guys. And so it's a treat. It's a real treat to be do this for everybody. You know, I'll open with people know OpenAI as the firm that builds ChatGPT. Mhmm. The product that they have in their pocket that comes with them every day to work, to personal lives. But the focus for today is OpenAI for enterprise. You guys lead …
Get the full transcript (15,286 words) + summary by email — free
One-time email with the complete transcript and AI summary of this episode. No account needed.
One email, no spam. We’ll also show you what SignalCast does.
Browse all BG2Pod with Brad Gerstner and Bill Gurley transcripts →
You just read a 3-minute summary of a 65-minute episode.
Get BG2Pod with Brad Gerstner and Bill Gurley summarized like this every Monday — plus up to 2 more podcasts, free.
Pick Your Podcasts — FreeKeep Reading
More from BG2Pod with Brad Gerstner and Bill Gurley
The SpaceX IPO, Fable 5, AI Capex Update & Market Check w/ Gavin Baker, Andrew Fox & Clark Tang | BG2
Jun 11 · 80 min
20VC (20 Minute VC)
20VC: Cohere Founder on How Cohere Compete with OpenAI and Anthropic $BNs | Why Counties Should Fund Their Own Models & the Need for Model Sovereignty | How Sam Altman Has Done a Disservice to AI with Nick Frosst
Sep 1
More from BG2Pod with Brad Gerstner and Bill Gurley
ChatGPT – The Super Assistant Era | BG2 Guest Interview
Mar 15 · 63 min
The AI Breakdown
Why OpenAI and Anthropic Are Becoming Consultants
May 5
More from BG2Pod with Brad Gerstner and Bill Gurley
We summarize every new episode. Want them in your inbox?
The SpaceX IPO, Fable 5, AI Capex Update & Market Check w/ Gavin Baker, Andrew Fox & Clark Tang | BG2
ChatGPT – The Super Assistant Era | BG2 Guest Interview
AI Enterprise - Databricks & Glean | BG2 Guest Interview
All things AI w @altcap @sama & @satyanadella. A Halloween Special. 🎃🔥BG2 w/ Brad Gerstner
AI Bubble, Stablecoin Boom, and Runnin' Down a Dream | BG2 w/ Bill Gurley and Brad Gerstner
Similar Episodes
Related episodes from other podcasts
20VC (20 Minute VC)
Sep 1
20VC: Cohere Founder on How Cohere Compete with OpenAI and Anthropic $BNs | Why Counties Should Fund Their Own Models & the Need for Model Sovereignty | How Sam Altman Has Done a Disservice to AI with Nick Frosst
The AI Breakdown
May 5
Why OpenAI and Anthropic Are Becoming Consultants
20VC (20 Minute VC)
Jul 4
20VC: Open Models vs Frontier Models: Who Actually Wins? | The $100,000 Token Budget Every Engineer Will Need | Why Forward-Deployed Engineers Are the Future of Enterprise AI with Clay Bavor, Co-Founder of Sierra
The AI Breakdown
Jul 1
Fable is Back: Here's What You Should Try First
Lenny's Podcast
Feb 12
“Engineers are becoming sorcerers” | The future of software development with OpenAI’s Sherwin Wu
Explore Related Topics
This podcast is featured in Best Investing Podcasts (2026) — ranked and reviewed with AI summaries.
Read this week's Health & Longevity Podcast Insights — cross-podcast analysis updated weekly.
You're clearly into BG2Pod with Brad Gerstner and Bill Gurley.
Every Monday, we deliver AI summaries of the latest episodes from BG2Pod with Brad Gerstner and Bill Gurley and 192+ other podcasts. Free for one show.
Start My Monday DigestNo credit card · Unsubscribe anytime