Ravi Gupta

I was sleeping in a lifeguard tower on a beach in Los Angeles. I had a bag of clothes, my laptop, and about three weeks of homelessness behind me. And I was floridly psychotic. I woke up in a Denny's one morning with no memory of how I got there. I was posting rap videos on Snapchat because I believed I was going to be famous. My dad found me at a Starbucks. My parents had been looking for me for days. That was the beginning of my fourth and final psychiatric hospitalization. I was 21. Three years later my mom found research on ketogenic diets for bipolar disorder. I went ketogenic. The mania stopped. I haven't been hospitalized since. I'm telling you this because somewhere, right now, there is a parent who has been told their child is treatment-resistant. That this is just going to be their life. There is hope in ketogenic and metabolic therapies.

I see something slightly different, but a convergence coming... Startups have to get profitable fast and lean. That incentivizes agents > humans, and offshore humans > onshore humans in waterfall fashion. Large companies w/ cash are most concerned about the existential risk to their underlying business model. So they're deploying capital to close the loop or transition from SaaS to RaaS as fast as possible, while shedding OPEX that wasn't contributing to the bottom line. But as token costs continue to increase, they are rapidly outpacing the cost of hiring human software engineers (esp offshore), but soon domestically held. So we're going to reach a tipping point where efficient companies deploy AI against their trickiest problems, or what they need running 24/7, and will have, ironically, cheaper labor, employing human beings for other tasks. The one mitigant will be capable, easy on-ramp local models. But their usage is still pretty frontier, not enterprise great enough to be relied upon for most companies.