A portrait of eight billion people
1. Key Themes
Theme 1: Simulation as the next foundational AI paradigm (digital twins of human behavior)
Simile is building a foundation model of human behavior
The core product is digital twins of real people used to predict reactions to decisions.
"He is the Co-founder and CEO of Simile, an applied AI company building a foundation model of human behavior. That involves creating digital twins of real people – like a writer or podcaster – and simulating how they might react to certain circumstances."
Simulation is positioned as co-equal with AGI
Park frames simulation as one of two pillars of advanced civilization, and the one he's drawn to.
"In any truly advanced civilization that gets depicted [in science fiction], you really see two pillars of scientific progress. One is some form of AGI, and the other a simulation. And I was always more drawn to the idea of simulation."
Capital markets are validating the thesis
"Simile has raised more than $300 million, most recently at a $2 billion valuation, to pursue them."
Theme 2: Enterprise decision-testing is the near-term commercial wedge
Test decisions on simulated customers before shipping them
Pricing, fees, and policy changes can be pre-tested against synthetic populations.
"CVS Health, for instance, can test a new pricing policy against thousands of simulated patients. A bank can model how customers will react before announcing a fee change."
Use cases extend well beyond corporate testing
The episode timestamps point to a wide surface area, including legal strategy, personal decisions, and investing.
"Use cases for simulation, from legal strategy to finding a partner" and "Simulation for investing"
Theme 3: Working backwards from desired outcomes (the "psychohistory" model)
The long-term goal is outcome-first, not prediction-first
Rather than asking Simile a question and receiving a forecast, the vision is to specify an end state and search for the path to it.
"Joon's ultimate goal is not to give Simile a question and have it make a prediction. It is to give a 'wicked' problem like climate change and have it simulate a superabundance of possible paths to help us find the best one."
Asimov's Foundation is the explicit inspiration
"He also pointed to Asimov's Foundation series as a core inspiration for Simile, and particularly the idea of psychohistory: giving a simulator a desired end state and working backwards to find the path there."
Theme 4: Competitive strategy in a world of ubiquitous simulation
If everything is predictable, where does edge come from?
Gabriele raises whether strategy itself gets commoditized.
"If simulation becomes as ubiquitous as Simile suggests it might, how do companies compete with one another? Is 'strategy,' as a broad discipline, solved?"
Remaining sources of advantage
"Joon pointed to a few thoughts on where edge might still live – namely in point of view, proprietary data, and compute."
Theme 5: Agentic, orchestrated "digital coworker" form factor
Persistent task-specific bots become the default for professional work
"I suspect this form factor – persistent, proactive, task-specific bots overseen by an orchestrator – will become the default for professional work. It's the closest thing to having a collection of digital coworkers."
2. Contrarian Perspectives
Simulation may concentrate advantage rather than democratize it
The conventional view is that powerful AI tools level the playing field. Gabriele argues the likely sources of edge favor incumbents.
"Joon pointed to a few thoughts on where edge might still live – namely in point of view, proprietary data, and compute. Now, this is just a thought experiment, of course. But two of those three presumably redound to the world's best-resourced companies, which is far from ideal."
Supporting reasoning: Proprietary data and compute are capital-intensive and scale with company size, so a world of ubiquitous simulation could entrench large players. Only "point of view" is plausibly accessible to small companies.
Strategy as a discipline might be "solved"
The provocative claim is that if outcomes can be simulated at scale, traditional strategic planning loses its premium.
"Is 'strategy,' as a broad discipline, solved?"
Supporting reasoning: If any firm can test thousands of scenarios against synthetic populations, differentiation through analysis erodes, shifting value to inputs (data, compute) and judgment (point of view).
Simulation, not AGI, as the more compelling frontier
Most AI attention centers on AGI. Park deliberately prioritizes simulation.
"One is some form of AGI, and the other a simulation. And I was always more drawn to the idea of simulation."
Supporting reasoning: The $300M+ raise at a $2B valuation suggests investors see simulation as a distinct, fundable category rather than a feature of general-purpose models.
3. Companies Identified
Simile
- Description: Applied AI company building a foundation model of human behavior via digital twins.
- Why mentioned: The subject of the episode and the primary investment-relevant company.
- Quotes: "Simile has raised more than $300 million, most recently at a $2 billion valuation, to pursue them."
- Description: Large U.S. healthcare and pharmacy company.
- Why mentioned: Example enterprise use case for pricing-policy simulation.
- Quotes: "CVS Health, for instance, can test a new pricing policy against thousands of simulated patients."
Grok (xAI) – Grok Bot
- Description: AI assistant/bot product.
- Why mentioned: The author calls it a strong example of the emerging agentic form factor.
- Quotes: "Grok Bot is really very good at this point."
Unnamed bank
- Description: Generic financial institution.
- Why mentioned: Illustrative customer for fee-change modeling.
- Quotes: "A bank can model how customers will react before announcing a fee change."
(Wiz, Moltbook, and HBO Max's The Rehearsal appear in the resource list without explanation, so their relevance can't be substantiated from the text.)
4. People Identified
- Description: Co-Founder and CEO of Simile; also a painter.
- Why mentioned: The guest and the source of nearly all insights.
- Quotes: "Joon Sung Park, Co-Founder and CEO of Simile" and "Joon, who happens to be a gifted painter in his own right."
- Description: Author/host of The Generalist.
- Why mentioned: Interviewer and author of the notebook commentary.
- Quotes: "I loved this line from Joon… I can sometimes forget to consider what it tells us about the author rather than the subject."
- Description: Economist known for agent-based models (segregation model).
- Why mentioned: Intellectual precursor to AI-driven simulation, per the episode timestamps.
- Quotes: "From Schelling's agent-based models to AI simulations"
Isaac Asimov
- Description: Science fiction author of the Foundation series.
- Why mentioned: Core inspiration for Simile's psychohistory-style approach.
- Quotes: "Asimov's Foundation series as a core inspiration for Simile."
- Description: Author of investigative nonfiction.
- Why mentioned: Author's current reading recommendation.
- Quotes: "Jon Ronson has a new book out, The Castle."
5. Operating Insights
Hire for "contradictory superpowers"
Look for candidates who combine two strengths that rarely coexist, such as rigor plus playfulness.
"When it comes to hiring, this is what Joon looks for: someone with two strengths that rarely go together. For example, a colleague with deep scientific seriousness and a very playful side."
Build "emotional rigor" into craft-heavy work
Whether in art, writing, or modeling people, deep understanding requires sustained connection with the subject.
"This was how he described the intensity one needs to build to paint a beautiful painting of someone. You need to build this deep, rigorous connection. This is a universal lesson, I think, and especially true in writing."
Pre-test consequential decisions on synthetic populations
Operators can de-risk pricing and fee changes before launch rather than learning from live customers.
"CVS Health, for instance, can test a new pricing policy against thousands of simulated patients."
6. Overlooked Insights
Simile uses simulation internally, including to model its own business
The timestamps hint at dogfooding, plus a surprising EV result and "siblings" of the core product.
"Using simulation internally at Simile" and "Surprising results from an EV simulation"
Trust and data sourcing are central, unresolved challenges
Several segments focus on how Simile gains customer trust and distinguishes public from hidden data, signaling these are the real moats and risks.
"Gaining trust" and "Public vs. hidden data and how Simile collects it"