Demystified · 19 entries
The jargon, decoded.
Plain-language explainers: what each AI term means, and why it matters to your work.
- →
Function calling / tool use
How a model stops narrating what ought to happen and actually sets it in motion. The bridge between language and action.
- →
Agentic workflow
A process where the agent picks the steps, not just executes them. The dividing line that changes how the workflow must be governed.
- →
Context window
How much input a model can see at once. Bigger is partly true, mostly misleading, and always expensive.
- →
Foundation model
The pre-trained base every AI feature sits on. Switching it is harder than vendors admit and easier than internal politics suggests.
- →
Embedding
Picture searching your enterprise knowledge base for “quarterly churn analysis” and retrieving a document titled “customer retention trends Q3” despite neither keyword appearing in the text.
- →
Multi-modal AI
Consider the most recent complex customer complaint your team handled: an angry voicemail, accompanied by a blurry product photo and a fragmented text description.
- →
Model Fine-tuning
Teaching an existing foundation model your domain and dialect: the shortcut to a custom model without building one from scratch.
- →
Inference
The moment a trained model is put to work, turning the expensive thing you built into answers, and into a running cost.
- →
Tokens
The small chunks of text models read and write, the unit that quietly determines AI cost, speed, and context limits.
- →
Hallucination
When a model states something fluent, confident, and entirely made up. Why it happens, and why it matters most in high-stakes work.
- →
Federated Learning
Training shared AI models on data that never leaves each owner's firewall: collective intelligence without surrendering proprietary data.
- →
Natural Language Processing (NLP)
The technology that lets machines read, interpret, and generate human language: the engine under chatbots, search, and summarization.
- →
Reinforcement Learning
Teaching an AI to make good decisions through trial, reward, and error: learning by experience rather than explicit instruction.
- →
Human-in-the-Loop versus Governance-in-the-Loop
Two ways to keep humans in charge of AI: reviewing individual decisions (human-in-the-loop) versus setting the rules the system runs under (governance-in-the-loop).
- →
Neural Network
A pattern-recognition system of layered, interconnected nodes, loosely modeled on the brain. The structure most modern AI is built on.
- →
Transformation Architecture
The Transformer: the 2017 "Attention Is All You Need" design that replaced older sequence models and made today's large language models possible.
- →
Telemetry of Work – What It Really Means
The digital exhaust of how work actually gets done: every click and handoff captured as data, like Formula 1 telemetry for the enterprise.
- →
Machine Learning
Software that learns patterns from data instead of following hand-written rules for every case.
- →
Agentic AI
AI that acts, not just answers.