1. AI agents remove all ‘friction’ in the economy
The scenario begins with AI agents undergoing a “capacity surge.” This has already happened. Citrini refers to Anthropic’s Cloud Codes and OpenAI’s Codex, both of which have wowed users with their performance in recent months.
Agents break into software-as-a-service companies like Monday.com, Zapier, and Asana, because they offer businesses an inexpensive way to perform tasks in-house, for example, managing databases and organizing workflows. This forces businesses like Oracle that rely on long-term contracts with customers to engage in a “race to the bottom” in terms of pricing.
Meanwhile, AI agents wreak havoc elsewhere. This scenario imagines that each consumer decides to use his or her own personal agent to conduct transactions and business. This completely sidelines companies that monetize the “frictions” in the economy, such as travel and property agencies that act as middlemen in processes like booking vacations or buying property.
Instead of using DoorDash, developers – and citizens – code their own food delivery apps, all of which compete, fragment the market, and destroy the margins of legacy businesses. Uber and other ride-sharing apps are also out of business. Instead of using Visa and MasterCard, AI agents decide to conduct all business in cryptocurrencies, as transaction costs are cheaper. This impacts traditional payment providers.
For Citrini, this is a logical endpoint for tireless AI agents who have the time and ability to optimize everything. “Habitual app loyalty, the entire basis of the business model, simply did not exist for one machine,” it writes.
In the real world, this scenario has caused shares of Uber, DoorDash, MasterCard, and American Express to decline this week.
2. mass white collar unemployment
Traditional narratives about progress imagine the latest technologies creating new jobs as they destroy others. This is not the case with AI.
Citrini writes, “AI is now a general intelligence that improves on tasks that humans will perform again. Displaced coders cannot simply move to “AI management” because AI is already capable of it.”
Instead, white-collar workers are en masse redeployed into unstable, gig-economy jobs — the author describes an imaginary friend of his laid off from Salesforce who drives for Uber. This leads to a reduction in wages in the area. Meanwhile, layoffs reduce consumer spending. Companies struggling with weak demand decide to invest in more AI, not workers.
“It’s a feedback loop with no natural breaks,” writes Citrini. The consequences are far-reaching when the wallets of the 10% of American workers, who account for 50% of consumer spending, are suddenly shut down.
3. Ripples in the broader economy
The scenario imagines that job losses and evictions at software companies will spill over into broader markets in two ways: through private loan defaults and the mortgage crisis.
Private credit firms, or lenders that are not banks, have been involved in the restructuring of many software businesses in recent years, making loans based on the projected annual revenues of those businesses in the future. Citrini gives the example of how Hellman & Friedman and Permira, an asset manager, took a software company called Zendesk private for $10.2 billion (£7.6 billion) in 2022. The acquisition included debt structured around the assumption that Zendesk’s revenues would be stable.
After AI agents, that assumption is no longer valid.
This leads to “the largest private credit software default” in history. This should be included in the software, Citrini writes, but it isn’t, because the capital on asset managers’ balance sheets consists of life insurance policies and “the savings of American families.”
Regulators have downgraded this software debt, which contributes to the 2027 crash.
Meanwhile, there is a mortgage crisis. White-collar workers no longer have white-collar jobs and are unable to pay their home loans. Citrini writes, “People borrowed against a future they could no longer believe in.”
4. downward spiral
All this makes the negative feedback cycle even worse.
The first-order cycle is that companies are laying off employees, which weakens demand and consumer spending, resulting in companies investing more in AI and laying off more employees.
The second-order spiral is that private credit turmoil and mortgage concerns mean markets have tightened, consumer confidence has been shaken, leading to more layoffs and more mortgage losses. “Each reinforces the other,” writes Citrini.
No fiscal policy tools exist to address this, because the crisis that is occurring in the real economy – job losses and suppressed wages and spending – is not the result of tight financial conditions that central banks can address, but rather the result of investment in AI, which makes “human intelligence less scarce and less valuable”.
The result is a crash driven by the mortgage markets in late 2027. This wipes out 57% of the S&P.
5. Capture of Silicon Valley and Ghost GDP
Citrini imagines that this crash will throw governments into a crisis they will be unable to manage.
“The system was not designed for this kind of crisis. The federal government’s revenue base is essentially a tax on human time. People work, companies pay them, the government takes a cut,” it writes.
“The government needs to transfer more money to households at precisely the time when it is collecting less money from them in taxes.”
However, AI companies are doing well. Big tech players building and selling AI models are earning handsomely. Since their companies hold large shares of markets, the economy looks great on paper.
Citrini has a term for this: ghost GDP, i.e. “production that appears in the national accounts but never circulates through the real economy”.
The social fabric has broken down and the Occupy Wall Street movement has closed the offices of AI companies for several weeks.
Citrini’s scenario ends with a warning: “This is the first time in history that the most productive assets in the economy have created fewer, not more, jobs. Nobody’s framework fits, because nobody was designed for a world where scarce inputs are abundant. So we have to build new frameworks. Whether we build them in time is the only question that matters.”
The impact of the Citrine scenario has surprised some commentators, including experts, who say AI tools are not yet able to implement it. Stephen Innes, managing partner of SPI Asset Management, says AI ideas have become market drivers.
“We have seen this market endure wars, sticky inflation, banking shocks, and tariff theatrics, yet one widely circulated Substack idea is enough to tip it over the edge,” he said.
