Risk and Reason

The episode contains numerous substantive insights about tenant screening, underwriting mechanics, and the counterintuitive mathematics of acceptable default rates. Brendan explains why zero eviction rates are economically suboptimal, walks through the NOI framework, discusses AI application constraints in regulated contexts, and contrasts different risk transfer mechanisms. However, there is moderate filler (personal anecdotes about the gecko ESA, the miniature pony service animal tangent, Mount Everest reference) that dilutes density in places.

“if you set those standards so high that no one breaks a lease or no one fails to pay back a mortgage, you would be making very, very few mortgages”

“the right amount of evictions is not zero”

Brendan brings genuinely counterintuitive framing - particularly the idea that zero eviction rates are undesirable and the detailed NOI/occupancy optimization model for rental assets. His critique of personal guarantors as a risk transfer mechanism is sharp and contrarian. The renter-side marketplace vision is forward-thinking. However, the core underwriting principles (credit scoring, fraud detection, speed vs. accuracy tradeoffs) are not novel, and much of the AI critique echoes established regulatory concerns.

“the right amount of evictions is not zero, they go white in the face”

“if you have AI doing any decision making...get an output that is like, yes, we should or should not rent to this person. Terrible idea. It is not going to be compliant”

Brendan is a director of product at a tenant screening company with substantive operating experience, having previously worked at Better (a real, high-profile mortgage startup) and a hedge fund. He demonstrates deep domain expertise in underwriting, risk modeling, and regulatory compliance. He is a practitioner, not a thought leader or career podcast guest. However, his seniority is somewhat modest - he is a director-level PM, not a C-level operator or founder with scale experience.

“I'm a director of product at a company called Findiggs that does tenant screening”

“went into mortgage startup, better...go from $0 to $7 billion back to near $0”

The episode includes concrete examples (Plaid integration, specific fraudulent documents, collections categories like Verizon and Progressive) and specific metrics (6 percentage point conversion boost for decisions within 12 hours, 50% fraud claim in Atlanta, 50% annual turnover for leasing agents). However, many claims lack hard numbers: the CEO of Graystar's 50% fraud claim is referenced skeptically without verification, the 'marginal' improvements of decision speed are discussed theoretically, and broader ROI calculations are absent. Some assertions remain vague.

“if you are giving a decision to someone who applied to an apartment within 12 hours, you get a six percentage point boost to the conversion”

“the average turnover for leasing agents at property managers is about 50% per year”

The host asks probing follow-up questions (e.g., about eviction costs, regulatory teeth, the trade-off between speed and fraud detection, AI compliance risks) and occasionally pushes back (e.g., requesting examples of niche support animals, questioning the 50% fraud claim). However, many softball moments exist: the host allows tangents about ESA animals and miniature ponies without redirecting, doesn't deeply challenge Brendan's claim that personal guarantors are universally bad, and doesn't ask for specific ROI data or customer outcomes. The conversation meanders productively but could be tighter.

“it's costly to evict to actually do the thing or what does that process look like?”

“could I have applicants pay for this insurance product that allows me to approve them even if they don't meet my credit standards because I know I'll be covered if they don't actually rent from or if they don't pay the rent”