The 1990 National Championship program won tonight 🏆👑 #StingEm🐝

If you made VC’s money as a founder Or even as a Head of Marketing or Head of BD, You will always have a job or ability to raise funding Often via referral from the VC’s that led the previous rounds for your past companies. No matter what your personal quirks are

NOBODY DESERVES IT MORE! See you NEXT SATURDAY 😎 buzz.gt/FBTIX #StingEm 🐝

This is amazing! A must read!

Everyone's obsessed with the last mile of AI—agents, copilots, the shiny stuff. But after 20 years building enterprise AI, I've seen where projects actually break down: the first mile. That's where data gets its identity—lineage, trust, context. Skip it, and you're feeding AI

Everyone focuses on AI models, algorithms, deployment. But most AI initiatives fail before any of that. The data feeding them isn't ready. We call it the First Mile problem. Talked about this with paulmuller on @Cloudera's AI Forecast. 🧵 🎧 Full episode: YouTube:

Hot take: your AI strategy isn't failing because of tools or talent. It's failing because your data architecture was never built to support it. We broke down 6 blind spots that keep stalling enterprise AI — and yes, we called the whitepaper "Unf*ck Your Data" because most data

Every enterprise architecture diagram shows clean generational transitions. Every enterprise I've worked inside has Oracle, Hadoop, Snowflake, and Iceberg all running at the same time. That gap is blocking AI readiness. We traced how four waves of data infrastructure

Tech used to be a world of "ors." This stack or that one. This vendor or that one. We're moving to a world of "ands." Spark AND Kafka AND Trino AND vector DBs. On-prem AND cloud. Legacy ML AND newer AI. Vendors who say "bring everything to us" are asking you to bet one stack

Most enterprise data teams are managing tenant environments with scripts, runbooks, and institutional knowledge. It works at 10 tenants. It breaks at 50. We built a sleek automated tenant management solution into NexusOne to solve this at the architecture level. Check out the

Every Iceberg adoption I've seen hits the same wall: the spec solves the format, but nobody solves for the operations: identity, multi-tenant provisioning, cross-engine policy enforcement, credential vending across cloud providers. Our Chief Architect wrote up exactly how we

AI agents today remind me of high school sex. Everyone's talking about it. Very few are actually doing it. And the ones who are? Not doing it well. Moving from demo to production takes 3 things: → Full data estate under one control plane (or agents hallucinate) → Treat

The platform you chose five years ago now owns you more than you own it. Apache Iceberg changes that. Full breakdown on what it means, and how to approach migration: open.substack.com/pub/nexuscogni…

Fifteen years ago enterprises called data "exhaust." Treated it like trash. That trash is now the raw material for every AI strategy in the market. The question of who owns it just became a business strategy question, not a compliance one.

Trino the engine is excellent. The operational model at multi-tenant scale is where things fall apart. Resource groups are a scheduling mechanism, not an isolation mechanism. We wrote up how to make it actually work: nexuscognitive.substack.com/p/trino-multi-…

My kid and I recently put the Death Star lego set together. It was hard. But it's significantly easier than making 85 open-source tools work together in an enterprise data estate. The tools aren't the problem. Making them work together is.

Only 7% of enterprises have genuinely AI-ready data. The other 93% are stuck in a permanent pilot trap: brilliant pilots that can't scale. Here's what real AI readiness actually requires: open.substack.com/pub/nexuscogni…

At SBFE's Annual Member Meeting today. Spoke on a panel about scaling AI responsibly in small business lending. NexusOne is proud to be the data and AI layer powering credit risk data for millions of small businesses across America.

When AI agents make bad decisions from incomplete data, the blast radius isn't one customer. It's the entire operation. And it compounds in hours, not months. The data estate has to be right before agents go live, not after.