The AI Podcast — Episode Summary

Episode: OpenAI Leadership Reshuffle, AI Unicorns, and White-Collar Work
Host: Jayden Schaefer
Date: January 23, 2026

Overview

In this episode, host Jayden Schaefer covers the shifting dynamics of the AI industry at the start of 2026, focusing on OpenAI’s leadership shuffle and enterprise challenges, the booming AI infrastructure sector, new benchmarks on AI’s impact on white-collar work, and an innovative attempt to automate calendar scheduling with AI agents. Jayden brings a conversational, insightful tone to news analysis, peppering anecdotes and notable insider perspectives throughout.

1. OpenAI Leadership Reshuffle and Enterprise Pressure

[00:30–04:45]

Key Discussion Points:

• Brett Zof Rejoins OpenAI:
  • Brett Zof, previously VP of Post Training Inference at OpenAI, returned to head enterprise sales after a stint founding Thinking Machine Labs, which had “raised a billion dollars…before any products were shipped.”
  • “The timing is pretty notable. OpenAI right now has a ton of pressure in the enterprise market, and I think this role is going to put him kind of at the center of their 2026 growth strategy.” (Jayden Schaefer, 02:35)
• Market Share Decline:
  • Despite launching ChatGPT Enterprise early, OpenAI’s enterprise LLM usage has “fallen roughly 50% in 2023 to about 27% by the end of 25”.
  • Anthropic leads the sector at 40%; Google’s Gemini is catching up.
  • Internal concern highlighted: “Sam Altman has apparently kind of flagged the fact that OpenAI is growing as a big concern inside of the company.” (03:40)
• Growth Focus:
  • CFO Sarah Fryer labeling enterprise as “a top priority this year”.
  • Partnership expansions, e.g., ServiceNow, signaling OpenAI’s push for “larger business customers”.

2. AI Infrastructure Unicorns: LiveKit & Infrax

[04:50–11:15]

LiveKit Hits Unicorn Status

• What is LiveKit?
  • Real-time voice/video infrastructure for AI, powers ChatGPT’s voice mode.
  • Raised $100 million at a $1 billion valuation: “The fact they were able to raise a hundred million dollars and of course they're fueling like OpenAI's voice mode…shows just how valuable this company and others like it will be.” (06:35)
• Customer Base: OpenAI, Xai, Salesforce, Tesla, emergency services, mental health providers.
• Industry Impact:
  • “As AI voice is becoming a lot more mainstream, the infrastructure layer is turning into a really valuable part of the stack.” (07:20)

Inference Startups: Infrax & Berkeley’s Ecosystem

• Focus on Inference:
  • Inference startups getting “massive seed rounds” as model deployment, speed, and cost efficiency take priority over basic research breakthroughs.
  • Infrax raised $150 million at an $800 million valuation; commercializes open source, especially VLLM.
• Industry Trend:
  • “Inference is where a lot of the real business value is actually being created. I mean the fact that they were able to raise 800 million I think goes to prove this.” (09:54)
• Academic Roots:
  • Projects like Radiac’s ARC and SGlang, “originated from UC Berkeley's AI research ecosystem,” showcase broader trends.

3. White-Collar Work and the Limits of Today’s AI

[11:20–15:15]

Key Insights:

• Apex Agents Benchmark:
  • New study from Merk (“Apex Agents”) gauges leading AI models’ performance on authentic white-collar tasks from consulting, law, investment banking.
  • Results: “Even the best performing models struggled to get more than about 25% of the questions right…Gemini 3 Flash was actually the leader here. It did 24%. Right behind it was GPT 5.2 at 23%. I think a lot of the others were closer to 18%.” (12:15)
• Where Models Struggle:
  • Main difficulty: “Not just kind of like reasoning in isolation, but…operating across multiple domains…pulling information from emails or documents or internal policies and collaboration tools…”
• Big Picture:
  • “AI is moving and improving a lot faster, but it's not yet ready to actually replace a lot of the high value professional roles inside of companies. So I think for now…the models look a lot…more like interns who need heavy supervision.” (14:10)
  • Quote: “You definitely can get these AI models to do one particular task very well and save you time on it, but it's not like it's going to go take over a full role yet. Yet, keyword, yet.” (14:41)

4. AI Agents and the Quest to Fix Scheduling

[15:20–17:50]

Block Kit and AI Calendar Agents:

• New Approach:
  • Former Sequoia partner Kazi Kimiji’s startup Block Kit just raised a $5 million seed round to automate scheduling via AI agents that “negotiate directly with each other to schedule meetings.”
  • “It's kind of like this meme…I recently saw on X which someone said, ‘I'll have my Claude contact your Claude and get back to you.’” (16:05)
• How it Works:
  • Agents coordinate on behalf of users’ calendars, taking into account “preferences, priorities, tone, flexibility…”
  • Integrates directly into email/Slack; can read user-specific rules like which meetings are movable or urgent.
  • Adopted by over 200 companies including Brex and prominent VC firms.
• Host’s Take:
  • “This is a really good example of how AIs are actually going to save us time, reshape a lot of daily workflows, and while they may not replace entire jobs today…they are speeding up and automating a lot of stuff that's happening.” (17:35)

5. Memorable Quotes & Notable Moments

• “The timing is pretty notable. OpenAI right now has a ton of pressure in the enterprise market, and I think this role is going to put him kind of at the center of their 2026 growth strategy.” (Jayden Schaefer, 02:35)
• “As AI voice is becoming a lot more mainstream, the infrastructure layer is turning into a really valuable part of the stack.” (07:20)
• “Even the best performing models struggled to get more than about 25% of the questions right…The biggest failure point was not just kind of like reasoning in isolation, but it was actually operating across multiple domains.” (12:15, 13:00)
• “The models look a lot less like autonomous workers and probably more like interns who need heavy supervision.” (14:10)
• “It's kind of like this meme... ‘I'll have my Claude contact your Claude and get back to you.’” (16:05)

6. Episode Timeline (Timestamps)

• 00:30–04:45: OpenAI leadership shuffle, enterprise competition, and priorities
• 04:50–07:45: LiveKit's unicorn funding and infrastructure importance
• 07:50–11:15: Inference sector investment surge; Infrax; Berkeley spin-offs
• 11:20–15:15: Merk’s benchmark: AI’s limits in actual white-collar work
• 15:20–17:50: AI agents automating scheduling; Block Kit’s approach and adoption

In Summary:
This episode spotlights where the real action and money flows in the evolving AI industry for 2026: shifting enterprise sands at OpenAI, huge bets on AI infrastructure and inference tooling, sobering data on AI’s actual workplace impact, and the practical magic of agent-to-agent collaboration. Jayden’s takeaways are candid, well-informed, and relatable for anyone tracking AI’s commercialization and day-to-day impact.