Travis Cronin.

A thin fluid layer of microscopic dark particles sandwiched between two transparent conductive glass sheets. Apply current, particles align, light passes — windows clear. Cut current, particles scatter — windows go opaque. The technology worked in controlled prototypes.

The fatal failure mode for automotive: if a car loses power, the windows go opaque. No automaker will ship a passive-fail-to-blackout system in a vehicle. The right fit was residential glass — but cost, complex wiring requirements, and long-term durability questions ultimately outweighed the consumer market opportunity. Shelved.

LearnedA working prototype isn't a product. Failure modes matter as much as features.

Born from pandemic-era health anxiety. The plan: bathroom and kitchen tiles with embedded UV-C lighting, sanitizing surfaces automatically with no user intervention.

Two things killed it. First, a UV-C component shortage delayed MVPs indefinitely. Second — more fundamentally — UV-C at 254 nm damages skin and eyes. Commercial systems require automated occupancy sensors and strictly unoccupied rooms; that's feasible in hospitals, not unsupervised in consumer homes. Add the cost, regulatory exposure, and secondary health hazards, and the math stopped working.

LearnedA safety profile incompatible with the use case isn't an engineering problem to solve — it's the wrong product.

The thesis: help ordinary people understand the contracts they're about to sign. We built it. I used it personally for almost a year on every contract that crossed my desk.

Two structural problems killed it for commercial release. Consumer legal review carries non-trivial liability exposure, and B2C distribution is brutal for a solo founder. A company called Ivo went on to build a more comprehensive B2B version of essentially this idea — and executed it beautifully. Hats off to them.

LearnedBuild for buyers, not users. And: regulatory risk in consumer legal isn't a tooling problem — it's structural.

Most pet owners can't name their closest emergency vet, let alone navigate to it under stress. The existing pet ambulance category covers only two narrow cases: inter-hospital critical care transfers, and immobile-animal lift assist (elderly owners with giant breeds, non-ambulatory situations). Genuine emergencies for the average household pet have no equivalent of 911.

Open questions before building it: persistent veterinarian shortage, ambiguous regulatory classification of animal ambulances as commercial transport vehicles, and zero government subsidy for animal care unlike human emergency services. The unit economics aren't obvious. Watching for the right co-founder with veterinary domain depth.

Data centers produce enormous waste heat, typically vented or used for low-grade district heating. Multi-stage flash distillation and similar thermal desalination methods can use that heat to convert seawater into fresh water — turning a disposal problem into a coastal water-supply solution.

The intersection sits at hyperscale compute, climate-stressed coastal cities, and water utility regulation. Whether the unit economics actually scale is the question I haven't answered. Worth more research time — and the right operator with experience in either hyperscale infrastructure or municipal water systems.

American "correctional facilities" rarely correct. The system is structured for punishment first; rehabilitation is treated as an afterthought, if it's resourced at all. Roughly four in ten incarcerated adults don't hold a high school diploma at intake — and the structural response to that gap is, for the most part, to ignore it.

The evidence on rehabilitation is unambiguous. The 2014 RAND Corporation meta-analysis found that people who participate in any correctional education are 43% less likely to return to prison. Recidivism falls to roughly 14% for those who earn an associate's degree while incarcerated, against a base rate above 50%. Every dollar invested in correctional education saves four to five in re-incarceration costs.

This is the cause I want to dedicate time to once the operating chapters close. Not a business — a contribution. Whether that means partnering with existing rehabilitation programs, funding work already underway, or building something alongside people closer to the system than I am, the specific path will depend on where I can do the most good. The intent is what's settled.