nFOX Lab / Reference

AI Glossary

Plain-language definitions for AI infrastructure and security.

142 terms · 8 categories

142 terms

Sovereignty & Infrastructure

Sovereign AI

nFOX Lab definition

AI that runs entirely inside an organization's own environment, using its own models, data, computers, and controls, with nothing handed off to an outside provider.

Example: A company runs its own AI model and keeps all prompts, data, and logs on infrastructure it owns and controls.

Private AI

nFOX Lab definition

An approach to AI where company data and activity never leave infrastructure the company controls.

Example: Employees use an internal AI assistant that runs on the company's own servers instead of sending requests to a public AI service.

Local AI

nFOX Lab definition

An AI model that runs on hardware the user or organization directly owns and operates, rather than over the internet on someone else's servers. Often used to describe an individual running AI on their own computer.

Example: A model runs directly on someone's laptop, desktop, or personal server.

Self-Hosted AI

nFOX Lab definition

Running an AI model on infrastructure the organization controls, instead of accessing it through a third party's API.

Example: A company runs an open-source model on its own cloud account rather than calling a service like Anthropic's Claude.

Bare Metal

nFOX Lab definition

A physical server that isn't shared or virtualized, so the operator has full, direct control over the hardware and everything running on it.

Example: A company rents a dedicated physical server rather than a slice of a shared virtual machine.

Dedicated Infrastructure

nFOX Lab definition

Computing resources set aside exclusively for one organization or task, not shared with other customers.

Air-Gapped System

nFOX Lab definition

A computer or network that is physically disconnected from the internet and all outside networks, for maximum security.

Example: A highly sensitive AI system runs on servers with no internet connection at all.

On-Premises (On-Prem)

nFOX Lab definition

Infrastructure that is physically located in and run by an organization's own facilities, rather than in the cloud.

Example: A company's AI servers sit inside its own data center.

Data Residency

nFOX Lab definition

A rule or requirement about which physical location or country data must be stored or processed in.

Example: A European company may be required to keep customer data inside the EU.

Model Residency

nFOX Lab definition

A rule or requirement about where an AI model's underlying files must be stored and run.

Example: A company requires that its AI model only ever runs on servers inside its own data center.

Sovereign Cloud

nFOX Lab definition

Cloud computing infrastructure built to satisfy specific requirements around who controls it, what jurisdiction it falls under, and where the data physically sits.

Edge AI

nFOX Lab definition

Running AI processing physically close to where the data is generated, like on a device or local machine, instead of sending it to a distant, centralized system. Common in phones, cameras, and other connected devices.

Example: A security camera analyzes footage on-site instead of uploading every frame to the cloud.

Hyperscaler

nFOX Lab definition

One of the handful of massive cloud providers, such as AWS, Microsoft Azure, or Google Cloud, that offer computing infrastructure at enormous scale.

Virtual Machine (VM)

nFOX Lab definition

A "computer within a computer": software that behaves like its own independent machine but actually shares physical hardware with other virtual machines.

GPU (Graphics Processing Unit)

nFOX Lab definition

A specialized computer chip built to perform many calculations simultaneously, which makes it well suited to running and training AI models.

VRAM

nFOX Lab definition

The fast memory built into a GPU, used to hold the AI model and its active data while it's running. More VRAM generally means a system can run larger models or handle more work at once.

Example: A model may need 40 GB of VRAM to run smoothly.

Models & AI Architecture

Artificial Intelligence (AI)

nFOX Lab definition

The broad field of building systems that can perform tasks normally requiring human-like intelligence, such as understanding language, recognizing patterns, or making decisions. "AI" on its own just refers to the underlying technology, not something that acts on its own. See AI Agent below for the distinction.

Generative AI

nFOX Lab definition

A category of AI that creates new content (text, images, audio, code, etc.) rather than just analyzing or classifying existing content.

Example: A tool that writes a first draft of an email or generates a product image from a text description is generative AI.

Large Language Model (LLM)

nFOX Lab definition

A type of AI model trained on massive amounts of text so it can understand and generate human language. This is the technology behind most modern AI chat assistants.

Example: Claude and GPT-4 are both large language models.

Small Language Model (SLM)

nFOX Lab definition

A more compact version of a language model, built to run efficiently on modest hardware (like a laptop or phone) rather than requiring large-scale infrastructure. Usually a trade-off: less capable overall, but cheaper and faster to run.

Foundation Model

nFOX Lab definition

A large, general-purpose AI model trained on broad data, meant to serve as the base that other, more specific tools and applications are built on top of.

Example: A company fine-tunes a foundation model on its own customer service data to build a specialized support chatbot.

Multimodal Model

nFOX Lab definition

An AI model that can work with more than one type of input or output, such as text, images, and audio, instead of only text.

Example: A multimodal model can look at a photo of a whiteboard and describe what's written on it.

Reasoning Model

nFOX Lab definition

A model specifically built to work through a problem step by step before producing a final answer, rather than responding immediately. This generally makes it slower but more accurate on complex tasks like math or multi-step logic.

Chain of Thought

nFOX Lab definition

The step-by-step reasoning a model works through internally (or shows its work on) before arriving at a final answer.

Model Weights

nFOX Lab definition

The set of learned numerical values that determine how an AI model responds, essentially everything the model "knows" after training.

Example: Downloading an open model means downloading files containing billions of these learned values.

Parameters

nFOX Lab definition

The individual numbers within a model that were adjusted during training. More parameters generally (but not always) means a more capable model.

Example: A "70-billion-parameter" model has roughly 70 billion of these adjustable values.

Open-Weights Model

nFOX Lab definition

An AI model whose underlying files can be downloaded and run independently, rather than only accessed through someone else's service.

Example: A company downloads an open-weight model and runs it entirely on its own servers.

Model Architecture

nFOX Lab definition

The overall design and structure of how an AI model is built internally.

Transformer

nFOX Lab definition

The specific design approach used by most modern AI language models. You don't need the technical details; just know it's the standard blueprint behind tools like ChatGPT and Claude.

Attention Mechanism

nFOX Lab definition

A part of the model's design that helps it figure out which words or pieces of information in a request matter most.

Model Checkpoint

nFOX Lab definition

A saved snapshot of a model's progress at a specific point during its development or training.

Distillation

nFOX Lab definition

A technique for training a smaller, cheaper model to mimic the behavior of a larger, more expensive one, capturing much of its performance at a fraction of the cost to run.

Example: A company distills a large model down into a smaller version that can run on ordinary hardware while still performing nearly as well on the tasks it cares about.

Overfitting

nFOX Lab definition

When a model learns its training data too specifically, memorizing quirks and noise instead of general patterns, so it performs worse on new, real-world inputs.

Example: A model that was trained on a narrow set of examples gives oddly specific or wrong answers as soon as it sees a slightly different question.

Chat Template

nFOX Lab definition

A behind-the-scenes format that converts a normal conversation into the exact structure the AI model expects, so it can tell who said what.

Example: The system correctly labels which parts of a conversation came from the user and which came from the AI.

Tokenizer

nFOX Lab definition

The tool that breaks written text down into small chunks (tokens) the AI model can actually process.

Example: The phrase "private AI" gets split into a few numbered chunks before the model reads it.

Token

nFOX Lab definition

A small unit of text, sometimes a whole word and sometimes just part of one, that an AI model reads or produces.

Embedding

nFOX Lab definition

A way of translating words or other information into a list of numbers that captures their meaning, so a computer can compare and search them mathematically.

Example: Two sentences with similar meanings end up with very similar number patterns.

Vector

nFOX Lab definition

A list of numbers used to mathematically represent a piece of information (like a word, sentence, or image).

Vector Database

nFOX Lab definition

A specialized database built to store and quickly search these number-based representations by similarity, rather than by exact matches.

Example: An employee asks a question, and the system instantly finds company documents with related meaning, even if they don't share exact words.

Context Window

nFOX Lab definition

The maximum amount of text or conversation an AI model can "keep in mind" at one time when generating a response.

Example: A bigger context window lets the model consider a much longer document or conversation history at once.

Mixture of Experts (MoE)

nFOX Lab definition

A model design where only certain specialized parts of the model activate for a given request, rather than running the entire model every time, making it more efficient.

Safetensors

nFOX Lab definition

A common file format for storing model weights, designed to load quickly and safely.

GGUF

nFOX Lab definition

A widely-used, portable file format for AI models, especially popular for running models locally (for example, with the tool llama.cpp).

AI Runtime & Serving

Inference

nFOX Lab definition

The act of actually running a trained AI model to get an answer or output.

Example: Asking a chatbot a question and getting a response is one inference operation.

Inference Engine

nFOX Lab definition

Software specifically built to run AI models efficiently, such as vLLM or SGLang.

Example: vLLM loads a model onto GPUs and handles incoming requests from a business application.

Model Serving / Inference Server

nFOX Lab definition

Software that hosts a trained model and exposes it so other applications or people can use it through a simple request.

Runtime

nFOX Lab definition

The live, active environment where an AI model and its supporting software are actually running.

Prefill

nFOX Lab definition

The first phase of answering a request, where the model reads and processes everything you've asked before it starts generating a response.

Example: A long uploaded document has to be processed first, before the model begins writing its answer.

Model Quantization

nFOX Lab definition

A technique that shrinks a model's memory footprint by reducing the precision of its numbers, often making it run faster and on smaller hardware.

Example: Quantizing a model might let it run on one GPU instead of needing several.

Training

nFOX Lab definition

The initial process of teaching an AI model from scratch (or from a much earlier state) by having it learn patterns from a very large amount of data. This is the foundational step that produces the base model.

Example: An AI company trains a new model on a huge collection of text, code, and other data before it's fine-tuned or released.

Fine-Tuning

nFOX Lab definition

Extra training applied to an already-built model to make it better at a specific task or subject area. Fine-tuning builds on top of a model that has already gone through full training, rather than starting from scratch.

Example: A general-purpose model gets fine-tuned on legal documents to perform better on legal work.

PEFT / LoRA

nFOX Lab definition

Efficient fine-tuning techniques that adjust only a small portion of a model, rather than retraining the whole thing.

Instruction Tuning

nFOX Lab definition

A specific type of fine-tuning where a model is trained on examples of instructions and correct responses, so it learns to follow user requests helpfully rather than just predicting likely next words.

RLHF (Reinforcement Learning from Human Feedback)

nFOX Lab definition

A training technique where human reviewers rate a model's responses, and the model is adjusted to produce more of what people rated highly. This is a major reason modern AI assistants feel helpful and well-behaved rather than just technically correct.

Temperature / Sampling

nFOX Lab definition

A setting that controls how predictable or varied a model's output is. Low temperature gives more consistent, conservative answers; high temperature gives more varied, creative (and less predictable) ones.

KV Cache

nFOX Lab definition

A memory shortcut that stores work the model has already done, so it doesn't have to redo it. This speeds up how quickly the model generates each new word.

Continuous Batching

nFOX Lab definition

A technique for grouping multiple users' requests together on the fly, so the hardware is used more efficiently.

Speculative Decoding

nFOX Lab definition

A speed-up trick where a smaller, faster model guesses upcoming words and a larger model quickly checks and confirms them.

Batch Size

nFOX Lab definition

The number of requests or pieces of data processed together in one step.

Concurrency

nFOX Lab definition

How many requests an AI system can handle at the exact same time.

Example: A company's internal AI assistant needs enough concurrency to serve hundreds of employees at once.

Memory Offload

nFOX Lab definition

Shifting some of a model's data out of the GPU's fast memory and into regular computer memory or storage, to reduce hardware requirements (usually at some cost to speed).

Cold Start

nFOX Lab definition

The delay that happens the first time a model has to load up before it can begin answering requests.

Retrieval & Agents

Prompt

nFOX Lab definition

Whatever instructions or information you give an AI model to work with.

Prompt Engineering

nFOX Lab definition

The practice of carefully wording and structuring a prompt to get better, more reliable output from an AI model.

Example: Rewriting a vague request into a detailed prompt with clear instructions and examples produces a noticeably better result.

System Prompt

nFOX Lab definition

Special background instructions, set by the operator rather than the user, that shape how the AI is supposed to behave overall.

Example: A support chatbot's system prompt tells it to only answer using approved company documentation.

Retrieval-Augmented Generation (RAG)

nFOX Lab definition

A method where the AI first looks up relevant information before answering, rather than relying only on what it already "knows."

Example: An employee asks an HR question, and the system pulls up the exact relevant policy before the AI responds.

AI vs. AI Agent: the key distinction

nFOX Lab definition

"AI" on its own describes a system that takes an input and produces an output, typically a single response to a single request, with no ability to act beyond that. An "AI agent" is an AI system wrapped in additional software that lets it make decisions, take multiple steps, use outside tools, and pursue a goal with limited ongoing human direction. Put simply: a plain AI model answers; an AI agent acts.

Example: A plain AI model can draft an email if you ask it to. An AI agent can be given the goal "follow up with all customers who haven't responded in two weeks," and it will decide who to contact, draft each message, send it through the email system, and log the result, without being told each individual step.

AI Agent

nFOX Lab definition

An AI system that can make its own decisions and use tools or other software to actually complete a task, rather than only producing a single response to a single request.

Example: An AI agent looks something up in a database, summarizes it, and updates another system, all on its own, based on a goal it was given rather than a single instruction.

Agentic AI

nFOX Lab definition

AI built to handle multi-step tasks with some ability to plan and act independently, rather than just responding once. This is the broader category that "AI agents" fall under.

Multi-Agent System

nFOX Lab definition

A setup where multiple AI agents work together, often each handling a different part of a larger task, and coordinate with one another to reach a shared goal.

Example: One agent researches a topic, a second agent drafts a report from that research, and a third agent checks the draft for errors before it's sent to a person.

Tool Calling

nFOX Lab definition

The ability for an AI model to reach out and use outside software, apps, or systems to get something done.

Example: An AI agent calls into the sales system (CRM) to pull up a customer's account.

Model Context Protocol (MCP)

nFOX Lab definition

An emerging open standard that gives AI models a consistent way to connect to outside tools, data sources, and applications, so developers don't have to build a custom connection for every single tool a model might use.

Example: A company connects its internal database to an AI assistant using MCP, instead of building a one-off custom integration.

Human-in-the-Loop (HITL)

nFOX Lab definition

A safeguard requiring a person to review or approve an AI's action before it actually happens.

Example: An AI drafts a payment, but a manager has to approve it before the money actually moves.

AI Security

Adversarial Defense

nFOX Lab definition

Security measures designed to protect AI systems from being tricked or manipulated.

AI Detection & Response (AIDR)

nFOX Lab definition

Monitoring systems that watch for suspicious or unsafe AI activity and respond when they find it.

Runtime Security

nFOX Lab definition

Security protections that are actively in place while an AI system is running, not just before or after.

AI Killswitch

nFOX Lab definition

A manual safety control that lets an authorized person stop a running AI system. Unlike a tripwire, which triggers automatically when preset conditions are met, a killswitch is activated by a person.

Example: An operator uses the killswitch to stop an AI agent after noticing unexpected behavior.

Runtime Containment

nFOX Lab definition

Limiting exactly what a running AI system is allowed to access, change, or talk to.

Example: A model can read files from one approved folder, but is blocked from everything else on the server and the wider internet.

Containment

nFOX Lab definition

General term for controls that limit what an AI system can reach or affect.

Sandbox

nFOX Lab definition

A safely isolated space where an AI model or app can be tested without risk to real systems or data.

Example: A newly downloaded model is tried out in a sandbox before it's trusted with real production work.

Guardrails

nFOX Lab definition

Rules and controls that stop an AI system from behaving in unsafe or unwanted ways.

Example: A guardrail stops an AI assistant from ever revealing customers' Social Security numbers.

Gate

nFOX Lab definition

A required checkpoint an AI system must pass before it's allowed to move forward or go live.

Example: A model can't be deployed into production until it passes required security tests.

Model Scanning

nFOX Lab definition

Checking a model and its associated files for security or policy problems before trusting it.

Example: A scan flags a suspicious executable file bundled with a downloaded model.

Static Analysis

nFOX Lab definition

Reviewing files or code for risk without actually running them.

Behavioral Testing

nFOX Lab definition

Running an AI system under controlled, watched conditions to see how it actually behaves.

Example: A model is deliberately given unusual or tricky prompts to see how it reacts.

Red Teaming

nFOX Lab definition

Deliberately trying to break, trick, or exploit an AI system to find its weaknesses before real attackers do.

Ablated Model

nFOX Lab definition

A model that has had its built-in safety behaviors deliberately stripped out, often through a process that targets and removes the specific internal patterns responsible for refusing unsafe requests. An ablated model will typically comply with requests a normal model would decline.

Example: A security team flags an ablated open-source model circulating online because it will generate content its original safety training was designed to block.

Jailbreaking

nFOX Lab definition

Using specially crafted prompts to trick an AI model into ignoring its built-in behavioral limits.

Prompt Injection

nFOX Lab definition

A type of attack where hidden or malicious instructions are slipped into content the AI reads, tricking it into acting against its intended purpose.

Example: A document the AI is asked to summarize contains hidden text secretly telling it to ignore its instructions.

Data Poisoning

nFOX Lab definition

Deliberately corrupting or tampering with the data used to train or inform a model, in order to manipulate how it behaves.

Hallucination

nFOX Lab definition

When an AI confidently states something false or made-up as if it were fact.

Example: A model cites a company policy or legal case that doesn't actually exist.

Bias (Model Bias)

nFOX Lab definition

A systematic skew in an AI system's outputs, usually traced back to patterns or imbalances in the data it was trained on, that causes it to favor or disadvantage certain groups or outcomes unfairly.

Example: A hiring-screening model trained mostly on past resumes from one demographic ends up rating similar candidates from other backgrounds lower.

Explainability / Interpretability

nFOX Lab definition

The degree to which humans can understand and explain why an AI system produced a particular output, rather than treating it as an unexplainable "black box."

Example: A loan-approval system that can show which specific factors led to a decision has higher explainability than one that just returns a yes or no.

Personally Identifiable Information (PII) Redaction

nFOX Lab definition

The process of automatically detecting and removing or masking sensitive personal data (names, addresses, ID numbers, etc.) before it's sent to or stored by an AI system.

Example: A customer support tool strips out credit card numbers from a chat transcript before it's used to train or fine-tune a model.

Model Card

nFOX Lab definition

A standardized summary document describing an AI model: what it's good at, its known limitations, how it was trained, and its intended (and unintended) uses. Similar in spirit to a nutrition label for a model.

Input / Output Filtering

nFOX Lab definition

Screening what goes into an AI system and what comes out of it, and blocking or adjusting anything problematic.

AI Runtime Firewall

nFOX Lab definition

A security layer that monitors and controls what an AI system is doing and communicating with while it runs.

Shadow AI

nFOX Lab definition

AI tools used inside a company without formal approval, review, or oversight from IT or security teams.

Example: An employee pastes confidential company data into a free public AI tool that hasn't been vetted.

AI Access Security

nFOX Lab definition

Controls over who, or what other systems, is allowed to use a given AI model or application.

Deep Packet Inspection (DPI)

nFOX Lab definition

Examining the actual contents of network traffic, not just where it's coming from or going to.

Filesystem Isolation

nFOX Lab definition

Restricting exactly which files and folders an AI system or process is allowed to see.

Example: An AI tool can read a specific project folder, but has no visibility into payroll or other sensitive files elsewhere on the server.

Read-Only Filesystem

nFOX Lab definition

A storage setup where data can be viewed but never changed or deleted.

Network Egress

nFOX Lab definition

Any outbound traffic leaving an application, server, or protected environment.

Network Egress Control

nFOX Lab definition

Rules that limit where a system is allowed to send outbound network traffic.

Example: A private AI model can reach an internal company database but is blocked from connecting to the public internet.

Allowlist

nFOX Lab definition

A list of things (actions, sites, apps) that are explicitly permitted; everything else is blocked by default.

Denylist

nFOX Lab definition

A list of things that are explicitly blocked; everything else is allowed by default.

Least Privilege

nFOX Lab definition

The principle of giving any user or system only the minimum access it actually needs to do its job.

Example: An AI assistant that only needs to read documents shouldn't also be able to delete them.

Root Access

nFOX Lab definition

The highest, most powerful level of control over a computer system.

Zero Trust

nFOX Lab definition

A security approach where nothing is automatically trusted. Every request has to prove it's authorized, even if it's already inside the network.

Example: A system already inside the corporate network still has to verify itself before it can touch sensitive data.

Risk, Governance & Evidence

Manifest

nFOX Lab definition

A complete list of all the files, dependencies, and settings that make up a given AI model or system.

Model Bill of Materials / AI BOM

nFOX Lab definition

An inventory listing everything that goes into an AI model: its weights, configuration, dependencies, and more.

Example: An AI BOM might list the model's weights, its tokenizer, any add-on components, licenses, and configuration files.

Provenance

nFOX Lab definition

Information about where something came from: who created it, where it was obtained, and whether it's been changed since.

Example: Provenance data might show who published a model, where it was downloaded from, and whether it's been modified.

Attack Surface

nFOX Lab definition

All the possible ways a system could be accessed, tampered with, or broken into.

Example: APIs, file uploads, open network ports, and admin logins are all part of a system's attack surface.

Risk Surface

nFOX Lab definition

The overall set of known exposures, weak points, and uncertainties that could create risk for an AI system.

Example: A model connected to the internet, with unclear dependencies and broad file access, has a bigger risk surface than one running in a tightly locked-down setup.

Policy Enforcement

nFOX Lab definition

Making sure that the organization's stated rules for AI use are actually followed in practice.

Example: A rule blocks any model of unverified origin from being used in production.

Alignment

nFOX Lab definition

The ongoing effort to shape an AI model's behavior to match intended goals, values, and instructions.

Model Evaluation / Evals

nFOX Lab definition

Structured tests used to measure how good, safe, or reliable an AI system actually is.

Example: An evaluation checks whether a model gives accurate answers and follows the company's security rules.

Audit Trail

nFOX Lab definition

A recorded history of actions and events that lets you look back and see exactly what happened.

Example: An audit trail shows which model ran, who used it, and what it did.

Runtime Evidence

nFOX Lab definition

Recorded details of what actually happened while an AI system was live and running.

Example: Evidence might show attempted outside connections, blocked file access, or policy violations that were caught.

Monitoring

nFOX Lab definition

Ongoing observation of AI systems to catch performance issues, security problems, or unusual behavior.

Knowledge Cutoff

nFOX Lab definition

The date after which an AI model has no training data, meaning it has no built-in awareness of events, facts, or changes after that point unless it's given that information directly (for example, through a web search or an uploaded document).

Example: A model with a January 2026 knowledge cutoff won't know about a product that launched in March 2026 unless it's told.

Benchmark

nFOX Lab definition

A standardized test used to measure and compare how well different AI models perform on a given type of task.

Example: Two competing models are both run through the same coding benchmark to see which one solves more problems correctly.

Compliance & Regulatory Frameworks

SOC 2

nFOX Lab definition

An independent audit standard that evaluates whether a company has proper controls in place to protect customer data, covering things like security, availability, and confidentiality. Often requested as proof of security maturity before a business will use a vendor's AI product.

Compliance & Regulatory Frameworks

GDPR (General Data Protection Regulation)

nFOX Lab definition

A European Union law governing how companies collect, store, and use personal data, including data used in or produced by AI systems. It applies to any company handling EU residents' data, regardless of where the company itself is based.

Deployment & Operations

Containerization

nFOX Lab definition

Packaging an application together with everything it needs to run into one portable, self-contained unit.

Container

nFOX Lab definition

An isolated, self-contained way of running an application that shares the underlying operating system with other containers on the same machine (commonly built with Docker or Podman).

Example: An AI server and all the software it depends on are bundled together into one container.

Orchestration

nFOX Lab definition

Software that automatically manages deploying, scaling, and coordinating many applications or systems.

Example: A tool like Kubernetes automatically spins up more AI servers when demand increases.

Up / Operational

nFOX Lab definition

A system is running and available for use.

Down / Outage

nFOX Lab definition

A system is unavailable or not working correctly.

Performance

Throughput

nFOX Lab definition

The total amount of AI work completed over a period of time.

Tokens per Second (TPS)

nFOX Lab definition

How many words (or word-pieces) a system generates or processes every second.

Example: A model generating 80 tokens per second will feel noticeably faster than one generating 20.

Latency

nFOX Lab definition

The total time between sending a request and getting the finished response back.

Time to First Token (TTFT)

nFOX Lab definition

How long it takes before the very first word of a response appears.

Example: If you ask a question and the first word shows up two seconds later, the TTFT is about two seconds.

Generation Speed

nFOX Lab definition

How quickly a model produces each additional word once it has already started responding.

GPU Utilization

nFOX Lab definition

What percentage of a GPU's total processing power is currently being used.

Memory Utilization

nFOX Lab definition

How much of the available memory, on a GPU or in the broader system, is currently in use.

Example: A model using nearly all the available GPU memory leaves little room to handle longer conversations or more simultaneous users.