Definition
Model drift
When a model's accuracy slips over time because the real-world data it sees no longer looks like the data it was trained on.
Learn the language of AI-native founders
Every week: the new terms, tools, and workflows founders are using, straight from the podcast.
Related terms in AI Fundamentals
ML modelA custom machine learning model trained on your own data, as distinct from a general-purpose LLM. More expensive and data-hungry to build, but harder to replicate because the training data is proprietary.ModelIn AI, the trained system that takes an input and produces an output, like a prediction, classification, or generated text.Model agnosticNot tied to one underlying AI model — able to route the same task across many models depending on cost, speed, or capability.Model CardA standardized documentation artifact that accompanies a machine learning model, reporting performance metrics broken down by demographic group and intersectional group. Used to detect and disclose bias before deployment.Model GatewayA routing layer that automatically sends tasks to the cheapest model capable of handling them, enabling cost control at scale without sacrificing output quality.Model routingSending different tasks to different underlying models based on which performs best on that task type. CEOfriend routes across Gemini, ChatGPT, Claude, and Grok.Modular GuardrailsBrand-safety constraints applied at the personality layer.MultimodalAble to process more than text, including images and audio. Dobbin reads layouts and critiques them against stated principles.