Hi! I'm the Founder of Meteor Dreams, a technology consulting company in Boston, MA. I write about building products and companies in stablecoins, AI/ML, blockchain, fintech, and more →
A language dies every 40 days. 38% of webpages from 2013 are already gone. We built seed vaults for biodiversity. Why haven't we built knowledge vaults, training AI models on endangered languages, traditional medicine, and martial arts lineages before the last practitioners are gone?
Some decisions can't wait for perfect information. Growing up in post-war Vietnam, I watched my father make hard calls that changed lives. Three days into Boston's biggest blizzard since 2015, I'm thinking about what it means to make the call on AI when the data isn't coming.
Most people focus on which AI model to use. But the quality of your results depends more on context than model choice. Here's what context actually is, why most of it is hidden from you, and what that means for getting better results.
A 2025 year-end analysis of the stablecoin chain explosion, tokenized deposits, and what the fragmentation means for cross-border payments.
Sun Tzu wrote about war, but he was really writing about uncertainty. How to win through positioning when you can't predict outcomes. That feels relevant right now as AI reshapes work.
Working in isolation started before COVID but the pandemic accelerated it. Through Web3, Crypto Winter, and now AI, I've watched us mistake distributed work for distributed humanity.
Retrieval gives agents information. Context gives them decision logic, constraints, and workflows. This deep dive explains why RAG cannot solve decision-making and why the context layer becomes essential infrastructure.
AI disruption is coming without coordinated solutions. Historical transitions had survivors and casualties. In the absence of a plan, individual agency and context ownership become survival strategies.
AI agents can execute actions, but they can't decide which action is correct. Finance exposes this gap more clearly than any domain. Here's why context, not compute, becomes the bottleneck.
A practitioner's warning to friends who aren't paying attention to AI. The transformation is already here, and most people won't see the punch until it lands.
