AI & Automation Glossary
Comprehensive AI and ML terminology dictionary. Clear definitions, business context, and practical examples for 65+ essential AI terms and concepts.
Accuracy
The proportion of correct predictions among total predictions. A basic classification metric that can be misleading for imbalanced datasets.
Accuracy
The proportion of correct predictions among total predictions, a basic metric for classification model evaluation.
Action
A step that an automated workflow performs, such as sending an email, updating a database, calling an API, or creating a document.
AI Agents
Autonomous AI systems that can perceive their environment, make decisions, and take actions to achieve specific goals. Unlike simple chatbots, agents can use tools, access external data, and execute multi-step tasks independently.
AI Alignment
The challenge of ensuring AI systems behave according to human intentions and values. Critical for making powerful AI systems safe, helpful, and beneficial.
AI Ethics
The principles and practices ensuring AI systems are developed and used in ways that are fair, transparent, and beneficial.
AI Readiness
An organisation's preparedness to successfully adopt and benefit from AI, including data, technology, talent, and cultural factors.
AI Strategy
A comprehensive plan defining how an organisation will adopt and leverage AI to achieve business objectives.
Airtable
A cloud-based platform combining spreadsheet simplicity with database power, enabling flexible data organisation and automation.
Amazon Bedrock
AWS's fully managed service for accessing foundation models from multiple providers including Anthropic, Meta, and Amazon through a unified API.
Anthropic
The AI safety company behind the Claude family of AI assistants, known for emphasising responsible AI development and constitutional AI.
API (Application Programming Interface)
A set of protocols and tools that allows different software applications to communicate with each other. In AI, APIs enable applications to access AI model capabilities without running the models locally.
API Documentation
Technical documentation describing how to effectively use and integrate with an API.
API Gateway
A server that acts as an entry point for APIs, handling request routing, authentication, rate limiting, and other cross-cutting concerns.
API Security
Practices and technologies for protecting APIs from threats including unauthorised access, data breaches, and abuse.
API Testing
Testing APIs to ensure they function correctly, handle errors gracefully, perform adequately, and are secure.
API Versioning
Strategies for managing changes to APIs while maintaining backwards compatibility for existing clients.
Attended Automation
Automation that runs on a user's workstation and works alongside humans, triggered and supervised by the user.
Attention Mechanism
A technique in neural networks that allows models to focus on relevant parts of the input when producing output. It's the core innovation behind transformer models and modern LLMs.
Augmentation
The process of enhancing AI model capabilities by connecting them to external data sources, tools, or knowledge bases. RAG (Retrieval Augmented Generation) is a prime example.
Authentication
The process of verifying the identity of a user, device, or system attempting to access a resource.
Authorisation
The process of determining what actions or resources an authenticated user or system is permitted to access.
AutoGen
Microsoft's framework for building multi-agent AI systems where multiple AI agents collaborate to solve complex tasks.
Automation Monitoring
Tracking automation performance, health, and outcomes through dashboards, alerts, and analytics to ensure reliable operation.
Automation ROI
The return on investment from automation initiatives, measuring the value delivered relative to the costs of implementation and operation.
Automation Testing
Validating that automated processes work correctly through systematic testing before and after deployment.
AWS Bedrock
Amazon's managed service for accessing foundation models from multiple providers through a unified API.
Azure OpenAI Service
Microsoft's hosting of OpenAI models on Azure, providing enterprise-grade security, compliance, and integration with Azure services.
Azure OpenAI Service
Microsoft Azure's enterprise service for accessing OpenAI models with Azure security, compliance, and regional data residency.
Backpropagation
The primary algorithm used to train neural networks by calculating gradients and adjusting weights to minimise errors. It propagates error signals backward through the network.
Batch Processing
Processing multiple items or transactions together as a group, typically scheduled for off-peak times.
Batch Processing
Processing data in groups or batches at scheduled intervals rather than immediately as it arrives.
Batching
Processing multiple requests or data points together in a single operation rather than one at a time. This improves throughput and efficiency in AI systems.
Benchmark
Standardised tests used to evaluate and compare AI model performance across specific tasks or capabilities.
BERT
Bidirectional Encoder Representations from Transformers - a landmark language model from Google that reads text in both directions to understand context.
Bias (AI)
Systematic errors in AI predictions caused by assumptions in the training data or algorithm. Can lead to unfair or inaccurate outputs for certain groups or scenarios.
Bidirectional Sync
Data synchronisation where changes in either connected system are reflected in the other, maintaining consistency in both directions.
Bot Management
Overseeing and coordinating software bots in an automation environment, including scheduling, monitoring, and maintenance.
Bubble
A visual programming platform for building full-featured web applications without writing code.
Business Case
A documented justification for an AI investment, including expected benefits, costs, risks, and strategic alignment.
Business Intelligence
Technologies and practices for collecting, integrating, analysing, and presenting business data. Enables data-driven decision making through dashboards, reports, and analytics.
Business Intelligence (BI)
Technologies and practices for collecting, integrating, analysing, and presenting business data to support better decision-making.
Business Process Automation
Using technology to automate repeatable business processes, reducing manual effort and increasing consistency. Broader than RPA, encompassing workflow systems and integrations.
Business Process Management (BPM)
A discipline for designing, executing, monitoring, and improving business processes to achieve organisational goals.
Automation Center of Excellence
A dedicated team or function that provides governance, standards, expertise, and support for enterprise automation initiatives.
Caching
Storing copies of data or computed results in faster storage to reduce latency and load on the original data source.
Center of Excellence (CoE)
A centralised team or framework that provides leadership, best practices, and governance for automation initiatives across an organisation.
Chain-of-Thought
A prompting technique that encourages AI models to show their reasoning step-by-step, leading to more accurate results on complex problems.
Change Management
The structured approach to transitioning individuals, teams, and organisations to a desired future state. Critical for AI adoption where people's roles and processes change.
Change Management
The structured approach to transitioning individuals, teams, and organisations from current state to a desired future state.
Chroma
A lightweight, open-source embedding database designed for AI applications, popular for its simplicity and ease of use.
Chroma
An open-source embedding database designed to be simple to use, particularly popular for development and prototyping AI applications.
Chunking
Breaking large documents or texts into smaller, manageable pieces for processing. Critical for RAG systems where documents must fit within context windows.
Churn Prediction
Using AI and analytics to identify customers likely to stop using a product or service, enabling proactive retention efforts.
Citizen Developer
Business users who create applications or automations using no-code/low-code platforms without traditional programming skills.
Citizen Development
Enabling non-IT employees to build applications and automations using low-code/no-code tools with appropriate governance.
Claude
Anthropic's family of AI assistants, known for being helpful, harmless, and honest. Claude models excel at analysis, writing, coding, and following nuanced instructions.
Cohere
An enterprise-focused AI company providing language models, embeddings, and search capabilities through a developer-friendly API.
Cohere
An enterprise AI company providing LLMs optimised for business applications, with strong focus on retrieval, classification, and embeddings.
Completion
The output text generated by a language model in response to a prompt. Also refers to the API endpoint type for generating text continuations.
Computer Vision
AI that enables computers to interpret and understand visual information from images and videos. Powers image recognition, object detection, and visual inspection.
Computer Vision
The field of AI that enables computers to interpret and understand visual information from images and videos.
Concept Drift
Changes in the relationship between input features and target variables over time, causing model predictions to become less accurate.
Confusion Matrix
A table showing predicted vs actual classifications, revealing true positives, false positives, true negatives, and false negatives. Essential for understanding model error patterns.
Confusion Matrix
A table showing the counts of correct and incorrect predictions for each class, revealing detailed model performance.
Connector
A pre-built integration component that enables applications to communicate with external systems, handling authentication and data formatting.
Context Window
The maximum amount of text (measured in tokens) that an LLM can process in a single request. This includes both the input prompt and the generated output.
Convolutional Neural Network
Neural network architecture designed for processing grid-like data such as images. Uses convolution operations to automatically learn spatial features and patterns.
CrewAI
A Python framework for orchestrating role-playing AI agents that work together as a crew to accomplish complex tasks.
Customer Experience
The overall perception customers have of their interactions with a company. AI and automation can dramatically improve CX through personalisation, speed, and consistency.
Customer Journey
The complete experience a customer has with a company, from initial awareness through purchase and beyond.
Customer Lifetime Value (CLV)
The predicted total revenue a customer will generate throughout their entire relationship with a business.
Customer Segmentation
The practice of dividing customers into groups based on common characteristics to enable targeted marketing and personalised experiences.
Customer Service Automation
Automating customer support through chatbots, self-service portals, ticket routing, and AI-assisted agent tools.
Dashboard
A visual display of key metrics and data points that provides at-a-glance understanding of business performance.
Data Augmentation
Techniques for artificially increasing training data by creating modified versions of existing data.
Data Catalog
A centralised inventory of data assets with metadata, enabling discovery, understanding, and governance of organisational data.
Data Cleaning
The process of detecting and correcting errors, inconsistencies, and inaccuracies in datasets to improve data quality.
Data Drift
Changes in the statistical properties of input data over time that can degrade machine learning model performance.
Data Governance
The framework of policies, processes, and standards for managing data assets. Ensures data is accurate, secure, compliant, and used appropriately across the organisation.
Data Labeling
The process of annotating data with labels or tags that machine learning models can learn from.
Data Labelling
The process of adding annotations or tags to data to create training datasets for supervised learning. Labels tell the model what output to predict for each input.
Data Lake
A storage repository holding vast amounts of raw data in native format until needed. Unlike warehouses, lakes store unstructured and semi-structured data without predefined schemas.
Data Lake
A centralised repository that stores raw data in its native format, enabling flexible analysis and machine learning workloads.
Data Lakehouse
An architecture combining data lake flexibility with data warehouse reliability and performance.
Data Lineage
Tracking the origin, movement, and transformation of data throughout its lifecycle in an organisation.
Data Lineage
The ability to track data from its origin through transformations to its final use, showing the complete data journey.
Data Pipeline
An automated sequence of data processing steps that moves and transforms data from sources to destinations. Essential for keeping AI systems and analytics fed with fresh data.
Data Privacy
Protecting personal and sensitive information from unauthorised access, use, and disclosure.
Data Quality
The measure of data fitness for its intended purpose. High-quality data is accurate, complete, consistent, timely, and valid. Critical for AI systems that learn from data.
Data Quality
The degree to which data is accurate, complete, consistent, timely, and fit for its intended use.
Data Security
Protecting data from unauthorised access, corruption, or theft through technical and administrative controls.
Data Sync
The process of maintaining consistency between data stored in different systems or locations.
Data Synchronisation
Keeping data consistent across multiple systems by automatically copying and updating data when changes occur. Essential for multi-system architectures.
Data Transformation
Converting data from one format, structure, or value system to another. Essential for integration when systems use different data models or formats.
Data Transformation
The process of converting data from one format or structure to another to meet the requirements of target systems or analysis.
Data Validation
The process of checking data against rules and constraints to ensure accuracy, completeness, and consistency.
Data Warehouse
A centralised repository that stores integrated data from multiple sources for reporting and analysis. Optimised for query performance rather than transaction processing.
Data Warehouse
A centralised repository optimised for analytical queries, storing structured data from multiple sources for business intelligence and reporting.
Databricks
A unified analytics platform combining data engineering, data science, and machine learning on a lakehouse architecture.
dbt (data build tool)
A transformation tool that enables analytics engineers to transform data using SQL, with software engineering best practices.
Decoder
The component of a transformer model that generates output sequences. GPT-style models are "decoder-only" architectures optimised for text generation.
Deep Learning
Machine learning using neural networks with multiple layers. Enables automatic feature learning and powers modern AI breakthroughs in vision, language, and more.
Diffusion Models
AI models that generate images by gradually removing noise from random patterns. Powers tools like DALL-E, Midjourney, and Stable Diffusion.
Digital Transformation
Integrating digital technology into all areas of business, fundamentally changing how organisations operate and deliver value. AI and automation are key enablers.
Digital Transformation
The integration of digital technology into all areas of business, fundamentally changing how organisations operate and deliver value.
Dimensionality
The number of features or dimensions in an embedding vector. Higher dimensionality can capture more nuance but requires more storage and compute.
Docker
A platform for containerising applications, essential for deploying AI models consistently across different environments.
Document Automation
Using AI and software to automatically create, process, extract data from, and manage documents without manual intervention.
Document Automation
Automatically creating, processing, routing, and managing documents using templates, data extraction, and workflow rules.
Email Automation
Automated sending, sorting, and responding to emails based on triggers, schedules, or content analysis. Includes marketing sequences, transactional emails, and inbox management.
Email Automation
Automatically sending, sorting, responding to, or processing emails based on triggers, rules, or AI-driven understanding.
Embeddings
Numerical vector representations of text, images, or other data that capture semantic meaning. Similar items have similar embeddings, enabling semantic search.
Encoder
The component of a transformer that processes input text into internal representations. BERT-style models are "encoder-only" and excel at understanding tasks.
ETL
A data integration process that extracts data from sources, transforms it to fit operational needs, and loads it into a destination system like a data warehouse.
ETL (Extract, Transform, Load)
A data integration process that extracts data from sources, transforms it for analysis, and loads it into a destination system.
Evaluation Metrics
Quantitative measures used to assess AI model performance, such as accuracy, precision, recall, F1 score, and perplexity.
Event-Driven Automation
Automation triggered by real-time events rather than schedules. Responds immediately when something happens - a form submission, database change, or API call.
Event-Driven Automation
Automations triggered in response to specific events or changes in systems, enabling real-time reactive workflows.
Exception Handling
Managing situations where automated processes encounter unexpected conditions or errors that prevent normal completion.
Exception Handling
Managing cases where automated processes encounter unexpected conditions, errors, or situations outside normal parameters.
F1 Score
The harmonic mean of precision and recall, providing a single metric that balances both. Useful when you need good performance on both false positives and false negatives.
F1 Score
The harmonic mean of precision and recall, providing a single metric that balances both concerns.
FAISS
Facebook AI Similarity Search, a library for efficient similarity search and clustering of dense vectors at scale.
Feature Engineering
The process of creating and selecting input variables (features) for machine learning models. Good features capture relevant patterns and improve model performance.
Few-Shot Learning
A technique where models learn to perform tasks from just a few examples provided in the prompt, without additional training.
Fine-Tuning
Adapting a pre-trained model to a specific task or domain by training it further on specialised data. Creates a new model variant.
Flowise
An open-source visual tool for building LLM flows and AI agents using a drag-and-drop interface built on LangChain.
Foundation Model
Large AI models trained on broad data that can be adapted to many downstream tasks. GPT-4, Claude, and BERT are examples that serve as the foundation for specific applications.
Foundation Model
Large AI models trained on broad data that can be adapted to a wide range of downstream tasks.
Function Calling
An LLM capability to output structured requests to external functions or APIs, enabling AI to take actions like searching databases or executing code.
AI Governance
The framework of policies, processes, and accountability structures that guide responsible AI development and use.
Generative AI
AI systems that create new content - text, images, code, audio, or video. Includes LLMs like GPT and Claude, and image generators like DALL-E and Midjourney.
Generative AI
AI systems that can create new content including text, images, audio, video, and code.
Google Vertex AI
Google Cloud's unified ML platform providing access to Google's AI models and tools for building AI applications.
Google Vertex AI
Google Cloud's unified AI platform for building, deploying, and scaling ML models, including access to Gemini and PaLM models.
GPT (Generative Pre-trained Transformer)
OpenAI's family of language models that generate human-like text. GPT-4 is currently the most capable version, excelling at reasoning, coding, and analysis.
Gradient Descent
The optimisation algorithm used to train neural networks by iteratively adjusting weights to minimise the loss function.
Gradio
A Python library for quickly creating web interfaces for machine learning models, particularly popular for AI demos.
Gradio
A Python library for quickly creating web interfaces for machine learning models with automatic UI generation.
GraphQL
A query language for APIs that lets clients request exactly the data they need. Alternative to REST that reduces over-fetching and enables flexible data retrieval.
GraphQL
A query language for APIs that allows clients to request exactly the data they need, developed by Facebook.
Ground Truth
The accurate, verified labels or outcomes used to train and evaluate machine learning models.
Grounding
Connecting AI model outputs to factual, verified information sources to reduce hallucinations and improve accuracy.
Guardrails
Safety mechanisms and constraints implemented to prevent AI systems from producing harmful, inappropriate, or off-topic outputs.
Hallucination
When an AI model generates plausible-sounding but factually incorrect or fabricated information. A major challenge in deploying AI for business.
Hugging Face
The leading platform for sharing and deploying machine learning models, datasets, and applications. Known for the Transformers library.
Hugging Face
The leading open-source AI platform providing model hosting, datasets, and the Transformers library for machine learning development.
Human-in-the-Loop
Automation design that includes human review, approval, or intervention points for handling exceptions or validating AI decisions.
Hyperautomation
An organisation-wide strategy to automate as many business and IT processes as possible using multiple technologies including AI, RPA, process mining, and low-code platforms.
Hyperautomation
A business-driven approach to rapidly identify, vet, and automate as many business processes as possible using multiple technologies.
Hyperparameters
Configuration settings that control the training process, such as learning rate, batch size, and number of epochs. Set before training begins.
Image Recognition
AI capability to identify and classify objects, people, text, and other content within images. Powers applications from photo organisation to quality inspection.
Implementation Partner
An external company that helps organisations plan, build, and deploy AI solutions.
In-Context Learning
The ability of LLMs to learn and adapt their behaviour based on examples and instructions provided in the prompt, without model updates.
Inference
Using a trained model to make predictions or generate outputs on new data. This is the "runtime" phase of AI, as opposed to training.
Instruction Tuning
Fine-tuning a model on examples of following instructions to improve its ability to understand and execute user requests.
Intelligent Automation
Combining RPA with AI capabilities like machine learning, NLP, and computer vision to automate processes that require judgment, learning, or understanding unstructured data.
Intelligent Automation
The combination of RPA with AI capabilities like machine learning and NLP to automate complex processes requiring judgement.
iPaaS
Cloud-based platforms that connect applications, data, and processes across cloud and on-premises environments. Examples include Zapier, Make, and Workato.
iPaaS
Integration Platform as a Service, cloud-based platforms for connecting applications and data across cloud and on-premise systems.
JSON
A lightweight data format for storing and exchanging data. Human-readable and easy to parse, JSON is the standard format for API requests and responses.
JSON
JavaScript Object Notation, a lightweight data interchange format that is easy for humans to read and machines to parse.
JWT
A compact, URL-safe token format for securely transmitting claims between parties. Used for authentication and authorisation in web applications and APIs.
Knowledge Base
A structured repository of information that AI systems can query. In RAG systems, this typically contains company documents, FAQs, and domain-specific content.
Knowledge Graph
A network of entities (people, places, concepts) and their relationships, enabling AI to reason about connections and context.
KPI
Quantifiable measures used to evaluate success in meeting objectives. For AI projects, KPIs track both technical performance and business outcomes.
KPI (Key Performance Indicator)
Measurable values that demonstrate how effectively an organisation is achieving key business objectives.
Kubernetes
An open-source container orchestration platform for automating deployment, scaling, and management of containerised applications.
LangChain
A popular open-source framework for building LLM applications. Provides abstractions for chains, agents, memory, and integrations with various AI services.
Langflow
A visual IDE for building and deploying LangChain applications through a flow-based interface with Python extensibility.
LangSmith
A platform from LangChain for debugging, testing, evaluating, and monitoring LLM applications.
Latency
The time delay between sending a request and receiving a response from an AI system. Critical for real-time applications.
Lead Scoring
A methodology for ranking prospects based on their perceived value and likelihood to convert, often enhanced by AI.
Llama
Meta's family of open-source large language models. Llama 2 and 3 offer strong performance that can be self-hosted without API costs.
LlamaIndex
A data framework for building LLM applications, specialising in connecting custom data to language models. Excellent for RAG applications.
LLM (Large Language Model)
AI models trained on vast amounts of text that can understand and generate human language. GPT-4, Claude, and Llama are leading examples.
LoRA (Low-Rank Adaptation)
An efficient fine-tuning technique that trains only a small number of additional parameters, dramatically reducing compute and storage requirements.
Low-Code
Development platforms that minimise hand-coding through visual interfaces while still allowing code customisation when needed. Bridges no-code and traditional development.
Make (Integromat)
A visual automation platform for connecting apps and designing complex workflows, formerly known as Integromat.
Marketing Automation
Software and strategies that automate marketing tasks including email campaigns, social media, lead nurturing, and customer journey orchestration.
Marketing Automation
Software and strategies that automate marketing tasks including email campaigns, lead nurturing, social media, and campaign analytics.
Memory (AI)
Systems that allow AI to retain and recall information across conversations. Can be short-term (within session) or long-term (across sessions).
Message Queue
A component that stores messages sent between applications, enabling asynchronous communication and decoupling between services.
Metadata
Data that describes other data, providing context about structure, meaning, origin, and usage.
Microservices
An architectural style where applications are composed of small, independent services that communicate over network protocols.
Middleware
Software that connects different applications or systems, handling communication, data transformation, and integration logic between components.
Middleware
Software that sits between applications, providing common services like messaging, authentication, and data transformation.
Mistral AI
A French AI company known for efficient, high-performance open-weight language models that compete with much larger models.
Mistral AI
A French AI company known for efficient, high-performance open-weight models that compete with larger proprietary models.
Mixture of Experts
Neural network architecture using multiple specialised "expert" subnetworks with a gating mechanism that routes inputs to the most relevant experts, enabling larger models with efficient compute.
MLflow
An open-source platform for managing the machine learning lifecycle, including experimentation, reproducibility, and deployment.
Model
A trained AI system that can make predictions or generate outputs. Models encode learned patterns from training data in their parameters.
Model Serving
The infrastructure and processes for deploying trained models to make predictions in production environments.
Multi-Agent Systems
Architectures where multiple AI agents collaborate, each with specialised roles, to accomplish complex tasks through coordination.
Multi-Modal
AI models that can process and generate multiple types of data - text, images, audio, and video. GPT-4V and Gemini are multi-modal.
Multimodal AI
AI systems that can process and generate multiple types of data such as text, images, audio, and video.
n8n
An open-source, self-hostable workflow automation platform with both visual builder and code capabilities.
Named Entity Recognition
NLP technique that identifies and classifies named entities in text - people, organisations, locations, dates, monetary values, and other specific information.
Natural Language Processing
The branch of AI focused on enabling computers to understand, interpret, and generate human language. Powers chatbots, translation, sentiment analysis, and more.
Natural Language Processing (NLP)
The field of AI focused on enabling computers to understand, interpret, and generate human language.
Net Promoter Score (NPS)
A metric measuring customer loyalty and satisfaction based on likelihood to recommend a company to others.
Neural Network
A computing system inspired by biological brains, consisting of interconnected nodes (neurons) that process information in layers.
NLP (Natural Language Processing)
The field of AI focused on enabling computers to understand, interpret, and generate human language.
No-Code
Platforms that allow building applications and automations without writing code, using visual interfaces and pre-built components.
Notion
An all-in-one workspace combining notes, docs, wikis, databases, and project management with AI capabilities.
OAuth
An authorisation framework that lets users grant limited access to their accounts on one service to another service, without sharing passwords.
Object Detection
Computer vision task that identifies and locates objects within images or video, drawing bounding boxes around each detected item and classifying what it is.
Ollama
A tool for running large language models locally on your own computer, making LLMs accessible without cloud APIs.
OpenAI
The AI research company behind GPT models, ChatGPT, and DALL-E. A leader in large language model development.
OpenAI Platform
The API and developer platform from OpenAI providing access to GPT models, DALL-E, Whisper, and embedding models for building AI applications.
OpenAPI Specification
A standard, language-agnostic format for describing REST APIs, enabling documentation, code generation, and testing.
Operational Efficiency
The ratio of output to input in business operations, often improved through AI automation and optimisation.
Orchestration
Coordinating multiple AI components, models, or agents to work together in a workflow. Managing data flow, error handling, and sequencing.
Overfitting
When a model learns training data too well, including noise and outliers, leading to poor performance on new data.
Parameters
The learned values (weights and biases) in a neural network that determine its behavior. LLMs have billions of parameters.
pgvector
A PostgreSQL extension that adds vector similarity search capabilities, enabling AI applications on existing PostgreSQL infrastructure.
Pilot Project
A small-scale preliminary project used to evaluate feasibility, test approaches, and learn before committing to full implementation.
Pinecone
A popular managed vector database optimised for AI applications. Known for ease of use and scalability.
Pipeline
A sequence of data processing or AI steps connected together, where each step's output feeds into the next.
Postman
A popular API development platform for designing, testing, documenting, and monitoring APIs.
Power Automate
Microsoft's cloud-based automation platform for creating workflows across Microsoft 365 and third-party applications.
Pre-training
Initial training phase where models learn general patterns from large datasets. Pre-trained models can then be fine-tuned for specific tasks with much less data.
Precision
Of all positive predictions, what proportion was actually positive. High precision means few false positives - when the model says "yes," it's usually right.
Precision
The proportion of true positive predictions among all positive predictions, measuring how reliable positive predictions are.
Process Assessment
Evaluating business processes to determine their suitability and priority for automation.
Process Mining
Analysing event logs from IT systems to discover, monitor, and improve business processes. Reveals how processes actually work versus how they're supposed to work.
Process Mining
Data-driven analysis of business processes using event logs to discover, monitor, and improve actual process execution.
Prompt
The input text or instructions given to an AI model to elicit a response. Quality prompts dramatically improve output quality.
Prompt Engineering
The practice of designing and optimising prompts to get better results from AI models. Combines art and science.
Pub/Sub (Publish-Subscribe)
A messaging pattern where senders (publishers) send messages to topics without knowledge of receivers (subscribers), who receive messages by subscribing to topics.
QLoRA (Quantized LoRA)
An even more efficient fine-tuning technique that combines quantisation with LoRA, enabling fine-tuning of large models on consumer hardware.
Quantization
Reducing the precision of model weights (e.g., from 32-bit to 4-bit) to decrease memory usage and increase speed with minimal quality loss.
Query
A request sent to an AI system or database to retrieve information or generate a response. In RAG, queries trigger retrieval from the knowledge base.
RAG (Retrieval Augmented Generation)
A technique that enhances LLM responses by first retrieving relevant information from a knowledge base, then using it to generate accurate, grounded answers.
Rate Limiting
Controlling the number of requests a client can make to an API within a specified time period to prevent abuse and ensure fair usage.
Real-Time Data
Data that is delivered and processed immediately or with minimal delay as it is generated.
Real-Time Processing
Processing data or transactions immediately as they occur, enabling instant responses and up-to-date information.
Recall
Of all actual positives, what proportion did the model identify. High recall means few false negatives - the model finds most of the positive cases.
Recall
The proportion of actual positive cases that were correctly identified, measuring how completely positives are found.
Recurrent Neural Network
Neural network designed to process sequential data by maintaining internal state. Used for time series, text, and other sequential tasks before transformers became dominant.
Reinforcement Learning
A machine learning paradigm where agents learn by interacting with an environment, receiving rewards or penalties for actions. Used in robotics, games, and optimisation.
Reinforcement Learning
Machine learning where an agent learns to make decisions by taking actions and receiving rewards or penalties.
Replicate
A platform for running machine learning models in the cloud via API, making it easy to deploy open-source models without managing infrastructure.
REST API
An architectural style for web APIs using standard HTTP methods. The most common way to build and consume web services, enabling system-to-system communication.
REST API
An architectural style for web APIs using HTTP methods to perform operations on resources, the most common approach for modern web services.
Retool
A low-code platform for building internal tools and admin panels by connecting to databases and APIs.
Retrieval
The process of finding and fetching relevant information from a database or knowledge base in response to a query.
RLHF (Reinforcement Learning from Human Feedback)
A technique to fine-tune AI models using human preferences, making outputs more helpful, harmless, and aligned with human values.
Robotic Process Automation
Software robots that automate repetitive, rule-based tasks by mimicking human interactions with digital systems. Works with existing applications without API integration.
Robotic Process Automation (RPA)
Software robots that automate repetitive, rule-based tasks by mimicking human interactions with digital systems.
ROI
A measure of profitability that compares the gain from an investment to its cost. For AI projects, ROI considers cost savings, revenue increases, and implementation costs.
ROI Measurement
The process of calculating return on investment by comparing the gains from an investment against its costs.
Rule Engine
Software that executes business rules to automate decisions, separating decision logic from application code.
Sales Automation
Automating sales tasks like lead assignment, follow-up sequences, proposal generation, and CRM updates to increase efficiency.
Sales Pipeline
A visual representation of where prospects are in the sales process, from initial contact to closed deal.
Scalability
The ability of a system or process to handle growing amounts of work or to be enlarged to accommodate growth.
Scheduled Automation
Automation that runs at predetermined times rather than in response to events. Used for batch processing, reports, maintenance tasks, and regular synchronisation.
Scheduled Automation
Automations triggered by time-based schedules rather than events, running at defined intervals or specific times.
Semantic Search
Search that understands meaning and intent rather than just matching keywords. Uses embeddings to find conceptually similar content.
Sentiment Analysis
NLP technique that determines the emotional tone of text - positive, negative, or neutral. Used for analysing customer feedback, social media, and reviews.
Sentiment Analysis
The use of NLP to identify and extract subjective information, determining whether text expresses positive, negative, or neutral sentiment.
Similarity Search
Finding items in a database that are most similar to a query, typically using vector distance calculations on embeddings.
Snowflake
A cloud-native data warehouse platform offering scalable storage and compute for analytics and AI workloads.
Streaming
Sending AI model output incrementally as it's generated rather than waiting for the complete response. Improves perceived latency.
Streamlit
A Python framework for quickly building and sharing web applications for machine learning and data science.
Streamlit
A Python framework for creating web applications and data dashboards quickly, popular for AI/ML demos and tools.
Structured Data
Data organised in a predefined format with clear schema, typically stored in databases with rows and columns.
Supervised Learning
A machine learning approach where models learn from labelled training data. The algorithm learns to map inputs to known outputs, enabling predictions on new, unseen data.
Supervised Learning
Machine learning where models learn from labeled training data to predict outcomes for new data.
Synthetic Data
Artificially generated data that mimics real data characteristics. Used when real data is scarce, sensitive, or expensive to obtain for AI training.
Synthetic Data
Artificially generated data that mimics real data characteristics while preserving privacy and enabling use cases where real data is scarce or sensitive.
Temperature
A parameter controlling randomness in AI outputs. Lower temperature (0-0.3) gives consistent, focused responses; higher (0.7-1.0) gives more creative, varied ones.
Text Classification
NLP task of assigning predefined categories to text. Used for spam detection, sentiment analysis, topic categorisation, and intent recognition.
Time to Value
The time between starting an investment and realising measurable benefits. Critical for AI projects where stakeholders expect results within reasonable timeframes.
Time-to-Value
The duration between starting an initiative and realising measurable business value from it.
Tokenization
The process of breaking text into smaller units (tokens) that AI models can process. Tokens might be words, subwords, or characters depending on the tokenizer.
Tokenization
The process of breaking text into smaller units (tokens) that AI models can process and understand.
Tokens
The basic units of text that LLMs process. Roughly 1 token = 4 characters or 0.75 words in English. Both input and output are measured in tokens.
Total Cost of Ownership
The complete cost of acquiring, deploying, and operating an AI system over its lifetime. Includes obvious costs like software and hidden costs like training and maintenance.
Total Cost of Ownership (TCO)
The complete cost of acquiring, operating, and maintaining a system over its entire lifecycle, including hidden costs.
Training
The process of teaching an AI model by exposing it to data and adjusting its parameters to minimise errors.
Training Data
The dataset used to train machine learning models. Training data teaches the model patterns and relationships it will apply to new, unseen data.
Transfer Learning
Applying knowledge learned from one task to a different but related task. This allows models to achieve good performance with less training data and compute.
Transfer Learning
A technique where a model trained on one task is adapted for a different but related task.
Transformer
The neural network architecture behind modern LLMs. Uses attention mechanisms to process sequences in parallel, enabling training on massive datasets.
Trigger
An event that initiates an automated workflow or action. Common triggers include form submissions, schedule times, data changes, emails, and webhooks.
Unattended Automation
Automation that runs independently on servers without human intervention, typically triggered by schedules or events.
Unstructured Data
Data without a predefined format or schema, such as text documents, images, audio, and video.
Unsupervised Learning
Machine learning where models find patterns in data without labelled examples. The algorithm discovers hidden structures, clusters, or relationships autonomously.
Unsupervised Learning
Machine learning where models find patterns in data without labeled examples or predefined outcomes.
Use Case
A specific scenario describing how AI will be used to solve a business problem or enable a capability.
Vector
A list of numbers representing data in multi-dimensional space. In AI, vectors (embeddings) encode semantic meaning of text or other data.
Vector Database
A specialised database optimised for storing and searching vector embeddings. Essential for RAG and semantic search applications.
Vendor Selection
The process of evaluating and choosing AI technology providers, platforms, or implementation partners.
Weaviate
An open-source vector database that combines vector search with traditional filtering, designed for AI applications.
Weaviate
An open-source vector database designed for AI applications, featuring built-in vectorisation, semantic search, and hybrid search capabilities.
Webhook
An HTTP callback that sends real-time data when events occur. Instead of polling for changes, systems push notifications to your endpoint automatically.
Weights
The numerical values in neural networks that are learned during training. They determine how strongly inputs influence outputs.
Weights & Biases
An ML platform for experiment tracking, model management, and collaboration in machine learning projects.
Weights & Biases (W&B)
A popular MLOps platform for experiment tracking, model visualisation, and team collaboration in machine learning projects.
Workflow
A defined sequence of automated tasks that accomplish a business process. AI workflows combine multiple AI and non-AI steps.
Workflow Engine
Software that executes and manages business workflows, routing tasks, handling approvals, and tracking progress through multi-step processes.
Workflow Template
A reusable, pre-configured workflow pattern that can be customised for specific use cases, accelerating automation development.