PerkyPet, Inc. ads

I spent a full day this week in a board meeting for the Bronx Science Foundation. Nine years I've been on that board, and somewhere around hour six I started thinking about hiring. Specifically, pedigree. The standard advice is to ignore it. I think that's only half right. Pedigree is a weak signal about the candidate. It can be a very strong signal about the person evaluating them. I unpacked why in today's Substack, and how the same thinking applies to something much broader: accountability. Read it here: https://lnkd.in/gb-_mgbt cc: Dara Huang MD, MMSc The Bronx High School of Science

I am the family member (Nephro) mentioned in this Substack post about the dangers of untrained, unqualified NP staffing in a community hospital under the umbrella of a major NY hospital system. The patient was self-admitted for a run-of-the-mill COPD exacerbation. He died 2.5 months later from an invasive fungal CNS infection after a prolonged ICU course complicated by human incompetency. There were multiple errors made, but primarily, these NPs have gone unchecked for too long and corporate healthcare is doubling down on AI to make it more efficient in delivering end of life outcomes faster. Which, by the way, was a late night phone call we got from an NP for a clotted dialysis machine?? Has anyone had any experience with physician whistleblowers? https://lnkd.in/eSUpjs8v cc: Tim Maliyil

Every AI-in-healthcare pitch I hear is about augmenting expert judgment. Better differentials. Faster reads. Decision support for the hardest calls. It is the wrong problem. My father died in an ICU where a positive test result sat in his chart for three weeks with no plan attached. Multiple credentialed clinicians read that chart. Each one assumed the result belonged to somebody else. Then a justification was finally written explaining the result away, and it was factually wrong. Wrong assay. Years-stale literature. Nobody caught it, because catching it would have meant checking the note against the evidence instead of against the seniority of the person who wrote it. He was not harmed by a hard call. He was harmed because an easy one never got made. Software is extraordinary at exactly this. It does not get tired at hour eleven. It does not defer to whoever signed the note. It does not assume somebody else has it. That is not augmentation. That is refusing to let a loop stay open. It is also the entire reason PerkyPet exists. https://lnkd.in/eUX2pXEM

Nobody builds anything alone. I've scaled companies, raised money, and taken risks that scared me. None of it happened by myself. Before any investor believed in me, my father did. My parents cleared the runway so I could take off. I wrote a tribute to him this week. It's the hardest thing I've ever published. Life is not fair. Sometimes in your favor, sometimes not. https://lnkd.in/ehNCmDU3

"Tell me, what is it you desire?" - Lucifer 😈 The devil is in the details. So is the demand. That distinction separates the companies that make it from the graveyard of AI products nobody missed. After 30 years of building, I have learned the founder's real job is not building. It is asking two questions until they become reflexive: What do people actually want? And what are they willing to do to get it? The second one is where the truth lives. Desire is easy. Effort is proof. We just saw it at PerkyPet. 80% of pet parents told us they would log their dog or cat's daily activity every day if it could add a year or more to their pet's life. That is not a wish. That is committed effort, and it is the only kind of signal you can build a company on. The asking never stops. Build less until you know what people want. Then never stop checking whether you still do. New piece on the discipline of asking, and the one survey number that stopped me cold: https://lnkd.in/gtccY48D

This was Harold before PerkyPet. 😉 Read more about him here: https://lnkd.in/deiTzYcY