Technologist, founder, and professor. I have built AI products since 2004, along with the teams behind them and the frameworks to think about both.
Co-founder and CTO at Tough Day. Adjunct professor at Cornell Tech, where I co-lead the Startup Studio.
I lead product and engineering at Tough Day, an AI workplace support platform that gives employees confidential help with professional challenges. My work covers agentic architecture, memory, and the practical side of putting AI to work inside companies.
I am an adjunct professor at Cornell Tech and co-lead the Startup Studio, where roughly 400 graduate students a year, in about 60 teams, build and validate new ventures in a single semester.
Venture building is the consulting work that starts where a workshop ends. The workshop produces a list of candidate services and products. Venture building turns the strongest ones into businesses, inside the company or spun out of it, each with a sponsor from the business, a customer as design partner, and a small founding team. Co-funding is on the table.
I have founded five companies and led one exit.
I taught in the Computer Science department at PUC-Rio in Rio de Janeiro. I spent nine years as an adjunct professor at Parsons School of Design in New York, teaching entrepreneurship. Today I am an adjunct professor at Cornell Tech, where I co-lead the Startup Studio for roughly 400 graduate students a year.
I write about the shape of agentic systems, how AI moves into companies, and lessons from building.
I publish essays and field notes on applied AI, agentic systems, and building AI products. New posts land every couple of weeks.
Open Substack →
Brazilians love a meme, and they jump on one faster than any other country. This week's meme was to ask ChatGPT to draw you as you would have looked in the 1980s, and the result is total culture slop.
Read on Substack →
Three MIT economists, including Nobel laureate Daron Acemoglu, published a Brookings paper making the case that most AI investment flows toward replacing human expertise, not expanding it. I agree with their direction and argue it should go further.
Read on Substack →