Perspectives on the rental economy.
Infrastructure, security, embedded finance, and the structural forces reshaping how money moves at the lease event.
When AI Agents Attack, Fragmented Infrastructure Loses
A $20 breach of McKinsey's Lilli platform in February 2026 revealed something every CTO had quietly assumed would never happen: every enterprise system was designed to stop humans, and AI agents are not humans. This series traces that vulnerability from first principles through the rental stack, and builds the case that only regulated financial infrastructure can serve as the spine AI agents require.
Every System Built Before AI Was Designed to Stop Humans
In February 2026, an AI agent breached McKinsey's Lilli platform for $20. It worked because every enterprise system was built for human-speed access. That assumption is now wrong across every SaaS platform running today.
The More Vendors You Add, the More Doors You Open
Every third-party integration is a trust boundary. The average enterprise runs 130+ SaaS applications. AI agents find the weakest link in the chain, and it is almost never at the primary vendor.
AI Agents Don't Buy Seats
SaaS was built around one assumption: charge per user. AI agents don't log in through GUIs, don't respect seat limits, and can do in minutes what took humans a week. This changes what enterprise software is actually worth.
The Tenant Who Never Existed
AI-generated synthetic identities now pass standard tenant screening. 93% of large operators report experiencing fraud. The pipeline generating it is more sophisticated than the infrastructure built to catch it.
What It Actually Takes to Build the Spine
Regulated infrastructure gets discussed loosely. Here is what it concretely means: the licences, the bank sponsorships, the carrier agreements, the bureau relationships, and why you need all of them simultaneously to matter.
The $20 Attack and What It Means for Your Rental Data
The McKinsey Lilli breach technique applies directly to PMC software stacks. Social insurance numbers, banking credentials, and underwriting AI live behind the same fragmented API architecture. The consequences are more severe.
The Infrastructure Being Built for AI Commerce Was Not Built for a Lease
Six protocol groups are racing to build the infrastructure layer for autonomous AI transactions. Every one of them is optimized for purchase commerce: a single buyer, a single seller, a confirmed price, a cleared payment. A residential lease is none of those things. This series maps exactly where agentic commerce infrastructure fails at the lease event, and what it will take to build something that does not.
The Protocol War Nobody in Rental Is Watching
Six groups of companies are racing to control the infrastructure layer for autonomous AI agents. They are designing protocols for buy-pay-confirm transactions. A rental lease is not that, and the difference has consequences.
Rent Is the Largest Unverified Transaction in Your Life
Rent represents 30-40 percent of household income for millions of renters. 96.5 percent of those payments produce no credit signal. The largest financial obligation most people carry is also the least verified.
The Rental Lease Is Not a Buy Button
The agentic commerce protocols being built today handle execution well. They do not handle governance. For most transactions, that is acceptable. For a residential lease, governance is the transaction.
Who Holds the Evidence When the Agent Makes a Mistake
3.6 million eviction filings happen in the US annually. Every one is a dispute about evidence. When AI agents execute lease transactions, the question of who holds the bilateral evidence record becomes the most consequential question in the transaction.
Why Regulated Infrastructure Gets More Valuable in an AI World
AI disrupts incumbents by compressing switching costs. Regulated infrastructure is the exception. AI agents cannot route around a KYC mandate, an insurance placement requirement, or a credit bureau furnisher relationship.
The Renter's Financial Identity Doesn't Travel
Mortgage history travels. Auto loan history travels. 96.5 percent of rent payments produce no credit signal. When AI agents begin executing lease transactions, they will present whatever financial identity exists. In rental, it is thin and stationary.