Definition
Intelligence boundary
Tarun's term for the undefined line between the data you hand a model and the understanding that model derives from it.
From the episode “Tokenomics Will Wreck Your Margins: The AI Model Routing Workflow That Cuts Spend”Hear it on the podcast1 episode
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
HyperparameterA setting chosen before training begins, like learning rate or model size, that shapes how a model learns.InferenceWhen a trained model is put to work on new inputs, making predictions or generating answers. It is what happens each time you use an AI product.Inference CostsThe computational expense generated when an AI model processes user queries. High inference costs signal heavy, active user engagement.Inference Engine (AI Context)In the context of Loman.ai, a system that draws contextual inferences from every customer interaction and stores them as a persistent diner profile. Rather than simply transcribing what the caller says, the inference engine captures implicit signals: background noise suggesting children or a social context, dietary patterns suggested by order history, regular timing patterns, and other behavioral cues. These inferences are used to personalize future interactions (surfacing likely orders, suggesting relevant upsells) and will eventually enable personalized outreach. The term distinguishes this layer of contextual reasoning from simple speech transcription or menu lookup - it is the layer that makes the AI agent feel like it knows the customer rather than merely recognizing their phone number.Intelligence layerThe analysis and decision-support layer built on top of the knowledge layer.InterpretabilityHow well people can understand why an AI model produced a particular decision or output.Intersectional BiasDiscriminatory model behavior that appears when demographic groups are combined — for example, a model that performs well for Black men and well for women, but poorly for Black women specifically. Requires testing at the intersection, not just the individual attribute level.ISO 42001International standard for AI management systems. The certification founders can pursue to signal trustworthy AI operations to enterprise clients and regulators.