Harvey's Margin Whiplash, Human Powered Agents, McLaren Rebrand, Insects > Humans? | Ben Thompson, Gagan Biyani, Alex Ratner, Peter Kalogiannis, Daniel Petkevich
September 22, 2026
AI Summary
5 min readThe AI Agent Economy Is Eating Everything—Including Margins, Mail, and Mosquito Morality
The episode opened with a stunner: Harvey, the legal AI startup, saw its gross margin swing from positive 50% to negative 50% in June as agent token use spiked 20-fold. That is the sound of the reasoning era arriving—and it hurts. Harvey was selling a dollar for 50 cents. But the story is more nuanced than a headline.
Harvey's Margin Whiplash and the Cost of the Agentic Era
The Bloomberg reporting that broke the story showed that Harvey's cost explosion came on the back of dramatically more product usage. Token consumption increased 20x this year, driven by a shift from simple one-shot queries to multi-step reasoning and then to full agentic workflows. A lawyer might now ask the system to "review this contract, then fan out across the firm to check all policies, MD files, and prior records." That generates vastly more tokens per task.
Continue reading the full summary in the app — free to try.
Read Full Summary →Free • No credit card required
Never miss an episode of TBPN
Get every new episode summarized in your inbox — free, ~5 minutes to read.
No spam. Unsubscribe anytime.
What you'll learn
- 1 (02:09) **Harvey's Margin Whiplash & Agentic Token Economics** - The panel breaks down Bloomberg's report on Harvey's gross margins swinging from 50% to negative 50% due to a 20x spike in agent token usage.
- 2 (07:22) **Legal Tech Margin Comps & The Cost of Reasoning** - The hosts compare Harvey's financial trajectory to established legal software firms and discuss the rapidly falling cost of frontier models.
- 3 (14:57) **Insects > Humans? The Effective Altruism Thought Experiment** - The group discusses a viral essay by "Bentham's Bulldog" arguing that the aggregate suffering of insects outweighs all human suffering.
- 4 (21:37) **A Concrete AI Doom Scenario: "The Old Way"** - One host narrates a detailed, cinematic scenario of an AI-orchestrated bioweapon attack and humanity's desperate, low-tech response.
- 5 (26:48) **Meta's Human-Powered Agents & The Data Collection Strategy** - The panel analyzes Meta's testing of human concierges for its new personal AI assistant, Muse, and the strategic implications.
- 6 (36:17) **The Frictionless Commerce Revolution & Sponsored Agentic Flows** - The panel explores how personal agents will eliminate profitable friction in consumer behavior, from unused flight credits to idle savings accounts.
- 7 (46:13) **McLaren's Rebrand & The Adult Toy Economy** - The panel reviews McLaren's controversial new logo and connects it to broader trends in adult consumer spending.
+ Full timestamped outline available in the app
Guests on this episode
Show Notes
- (01:32) - Harvey's Margin Whiplash
- (12:54) - Insects > Humans?
- (26:44) - Human Powered Agents
- (45:59) - McLaren Rebrand
- (50:27) - Toy Sales Growth Driven by Adults
- (56:12) - Ben Thompson, founder of the technology analysis publication Stratechery, discusses how AI agents, personal assistants, and future robotics can eliminate tedious real-world tasks. He also explores consumer behavior, arguing that people value relief from everyday friction but still enjoy activities like shopping, while Amazon’s logistics infrastructure gives it a durable advantage over Shopify and other competitors.
- (01:38:16) - Gagan Biyani, CEO of Horowitz Andreessen Academy, discusses the new for-profit San Francisco school for young builders, backed by $42 million in funding. He outlines its tuition-free founding class of 50 students, one-year immersive program beginning in fall 2027, and mission to prepare students for entrepreneurship or impactful roles at established companies.
- (01:45:21) - Alex Ratner discusses Snorkel AI’s $350 million funding round and the shift from high-volume training data to precise, specialized data targeting model weaknesses. The Snorkel AI co-founder argues that recursive self-improvement will continue to require human expertise, real-world input, and sophisticated combinations of people and AI.
- (01:53:24) - Peter Kalogiannis discusses Swarm Aero’s Gamera, a low-cost autonomous aircraft designed for long-range missions and large payloads. The co-founder and CEO explains the company’s integrated approach to aircraft manufacturing, scalable production, and swarm autonomy, with a focus on potential operations in the Pacific.
- (01:57:55) - Daniel Petkevich discusses his entrepreneurial career, including founding Fair Square Medicare and Trim, and his current car show, *Cars and Capitalists*. He also shares predictions about the Tesla Roadster and describes building an open-source quadruped robot dog.
TBPN is made possible by:
Ramp - https://ramp.com
Public - https://public.com
Cisco - https://www.cisco.com
Console - https://www.console.com
CrowdStrike - https://www.crowdstrike.com
Figma - https://www.figma.com
MongoDB - https://www.mongodb.com
NYSE - https://www.nyse.com
Railway - https://railway.com
Shopify - https://www.shopify.com
Codex - http://openAI.com/codex
Follow TBPN:
https://TBPN.com
https://x.com/tbpn
https://open.spotify.com/show/2L6WMqY3GUPCGBD0dX6p00?si=674252d53acf4231
https://podcasts.apple.com/us/podcast/tbpn/id1772360235
https://www.youtube.com/@TBPNLive
More from this podcast
TBPN →