Valuate the Model, Don’t Just Evaluate It
Gooder AI CEO Eric Siegel on reporting what a model is worth rather than how it scores, the one extra step on the same test data, and why projects stall.
The Data Itself Is the First Prompt in Vibe Analytics
Plotly’s Domenic Ravita on vibe analytics: no first prompt, the dataset is the prompt, and why he will not guess how much AI cuts the data cleanup.
Extend Data Governance Into Models, Then Into Agents
BARC’s Kevin Petrie on extending data governance into models and agents, why the top AI control is still a human, and where data quality ranks.
From Clicks to Conversions: Pay Only for Measured Outcomes
Matthew Swanson of Motion Enterprises on agents that replace clicking with conversation, and pricing that charges only for measurable gains in a KPI.
When Attackers Have AI, Verify the Person Each Time
Kris Bondi of Mimoto on why anomaly detection misses a stolen account, and what continuous person-level validation catches once a deepfake is inside.
Bridging Data Science and Generative AI
Domenic Ravita of Plotly on where data science and generative AI meet: custom data applications for operational decisions that nobody sells off the shelf.
Media Metadata Is a Signal to Detect Real From Fake
Allan McLennan of PADEM Media Group on metadata as the only proof a piece of content is yours, and on what broadcasters build to prove what is genuine.
AI Is a Tool for Augmenting People, Not Replacing Them
Sean White on AI as a tool that augments people, owning your own intelligence on premises, and the no surprises rule for wearable data.
Judge a Model on Cost and Latency, Not Just Accuracy
Ivan Lee of Datasaur on judging an AI model in production: unit cost first, then latency, then quality, and no single model wins.
Synthetic Data Has a Place, but It Is Not the Bias Panacea
Matt Maccaux of Google Cloud on whether you can generate your way out of biased training data, and why he leans on humans in the loop instead.
Bring the AI to Your Data, Not Your Data to the Cloud
Mark Heynen of Knapsack on why the fix for AI data risk is architectural: run the model where the data already sits instead of uploading it.