My daughter is three years old. She broke her toy guitar and wants a new one. Instead of another toy guitar, I bought her a real Ukulele. It’s small, lightweight, and perfect for a small girl. Besides, it sounds way better than the annoying toy guitar. She loved it. But I’m the one who played …
I Vibe-Coded an App in Just Two Hours (And Regretted It the Next Day) Read More »
When I write, I picture a cabin in the mountains. I have never lived in one. The picture is enough. It quiets my mind and lets it move. That is the first thing I learned about writing. The room is mostly in my head. For years I thought I needed what writers are told to …
The room where I write, and rebuild it every morning · Hanh D. Brown Read More »
Most production AI agents still send every LLM call to the same expensive frontier model. Classification steps, simple tool calls, progress checks, and hard reasoning all hit the same endpoint. The result is unnecessary cost and latency. NVIDIA NeMo Switchyard solves this. It is an open-source routing layer (proxy + library) that sits between your …
Switchyard: NVIDIA’s Open Source Routing Library Read More »
Short answer What is the real test of AI in the next five years? The real test of AI in the next five years is whether it reaches the people who need it most. The frontier capability is being absorbed by enterprises and hedge funds that already had access to capability. The audience that would …
Reaching the People Who Need It Most · Hanh D. Brown Read More »
ContentsReal-world data at continuous global scalePrecipitation forecasting at breakthrough accuracy Real-world data at continuous global scale WeatherNext 3’s biggest leap forward is what it learns from. Most AI weather models, including WeatherNext 2, are trained on data from numerical weather prediction (NWP) models. Although useful, NWP models are complex, supercomputer-driven physics simulations that carry a …
Short answer Which workers does AI hit first, and what does the forecast actually say? White-collar work goes first. Blue-collar work goes later. The industrial-era story has the order wrong. Anything that lives in bits gets automated now. Anything that needs hands on atoms takes longer. Abundance and unrest arrive together in three to seven …
Abundance and Unrest Hit the Middle of the Labor Market · Hanh D. Brown Read More »
Every major inference framework shipped prefill-decode disaggregation this year. NVIDIA built it into Dynamo. SGLang made it the default for large-scale deployments. vLLM added a KV connector API to support it natively. The consensus is forming fast: split your prefill and decode onto separate GPU pools, and throughput improves. The consensus is wrong for most …
AI’s Economic Future: Slow Upside, Fast Downside, Weak Links. A chain breaks at one link and strengthens one at a time. That asymmetry is why AI’s economic upside arrives slowly and its downside arrives fast. ai-economics ai-policy ai-careers A chain is only as strong as its weakest link. Artificial Intelligence (AI) makes the strong links …
Slow Upside, Fast Downside, Weak Links · Hanh D. Brown Read More »
We ran an experiment: three small datasets, one AI model, and the questions a business team asks in a normal week — what’s our average delivery time, which region is our best performer, how many athletes are in this file. Then we added a review pass. We handed the model its own answer back and …
I Asked ChatGPT to Analyze 3 Datasets. It Made the Same Mistakes Every Time Read More »
AI Is Not a Faster Chip. It Is a Reinvention of Computing.. AI is not a faster chip. It is a wholesale reinvention of how computers work, and the sentence the chip-and-forecast discussion does not say is the one that matters most. ai-hardware ai-economics ai-policy For sixty years, software was written by humans typing rules …
AI Is Not a Faster Chip. It Is a Reinvention of Computing. · Hanh D. Brown Read More »
