Tools & Usage
48
The terms you meet when actually using and connecting AI. Agents, APIs, open-source models, running locally, multimodal, plugins: what you need to know to pick and combine AI services.
Level
A/B Test
A test that shows two versions to real people and compares results
AI AgentAn AI that takes a goal and works through the steps on its own
AI AssistantA tool that asks, organizes, and uses linked tools to help with work
AI Coding AssistantA tool that helps out beside you while you write code
AI NotebookA workspace that reads only what you put in and answers from that
AI SearchA search that reads the web for you and hands back a written answer
APIThe fixed format two programs use to trade requests and answers
API KeyThe ID you present when you call an API
Browser AIAI that runs inside a web page with nothing installed
CachingKeeping a result you made or fetched once close by to reuse it
CheckpointA saved snapshot of a model's state during or after training
ComputeThe total amount of calculation it takes to train or run a model
Computer-Use AgentAI that watches the screen and moves the mouse and keyboard for you
Conversation MemoryRe-reading past conversation every time to keep it going
Document Q&AAsk questions about one document and get answers pulled from it
Function CallingA model requesting a tool in a fixed, structured form
GuardrailsThe safety fence set up in advance to keep AI from crossing a line
Human in the LoopPutting a person's judgment at key points in an AI workflow
LatencyThe time between sending a request and getting an answer back
Local LLMRunning an AI model directly on your own computer
Model CardThe document that spells out what a model can and can't do
Model HubA place where open models are collected for anyone to grab
Model ServingPutting a finished model to work answering everyone's requests
Model SizeHow many numbers a model is carrying
Multi-AgentSeveral AIs splitting up roles and passing work between them
MultimodalAI that handles text, images, and sound together
No-Code AIBuilding an AI by clicking through a screen, with no code
On-Device AIAI that runs directly on the device itself
Open WeightsA model that hands out only the finished trained numbers
Open-Source ModelAn AI model that shares how it was made, not just the result
OrchestrationWiring several steps and tools together into one flow
PersonaThe role and tone you assign an AI to speak as
PluginAn attachment you snap onto a base tool to add a capability
PromptThe written request you hand an AI, spelling out what to do
QuantizationCoarsening the scale numbers are written on to make a model lighter
RAGRetrieval-Augmented GenerationFetching relevant material before answering, then writing from it
Rate LimitA ceiling on how much you can use within a set stretch of time
Real-Time InferenceFinishing a judgment on incoming input within a short, fixed window
SDKA development kit that helps build programs and connect services
Semantic SearchSearch that finds things close in meaning, not matching letters
Small Language ModelA language model shrunk down to run on ordinary devices
StreamingSending each piece of an answer the moment it's ready
System PromptThe operating rules a service lays down before a conversation starts
Tool UseChoosing the right outside means and using it when needed
Vector DatabaseA store that keeps sets of numbers so the nearest ones surface fast
Voice AssistantAn AI you speak commands to, and hear answers from
VRAMThe workspace where a graphics card lays out its numbers to compute
Workflow AutomationChaining several steps so they run through on their own