Glossary
Plain-English explanations of key terms across Lean methodology, artificial intelligence, and the Lean AI approach that underpins everything we do at Boxtree.
What is Lean AI?
Lean AI is the deliberate integration of Lean methodology with AI implementation, using the proven diagnostic discipline of Lean to identify where AI can eliminate waste, release capacity and deliver measurable operational improvement. It is the foundational methodology behind everything Boxtree does.
Lean AI
The practice of applying Lean methodology to identify and prioritise AI implementation opportunities. Rather than deploying AI speculatively, Lean AI starts by mapping and quantifying waste, then applies the most appropriate AI tools to eliminate it. The result is purposeful, measurable AI improvement rather than technology for its own sake.
Problem-First Approach
The principle of beginning with operational challenges rather than available technology. Boxtree always identifies the problem before selecting any AI solution, ensuring every tool deployed is the right one for the specific challenge, not the most convenient or fashionable one available.
Lean AI Diagnostic
The Discovery stage of a Boxtree engagement, applying Lean's process-mapping and waste-identification techniques to an organisation's operations to quantify AI improvement opportunities and rank them by expected return. Combines value stream mapping, Gemba observation and data analysis.
Cost-Neutral AI
An AI implementation engagement where the measurable financial return in Year 1 equals or exceeds the total cost of our support. Boxtree guarantees cost-neutrality for every engagement by building a rigorous business case before any work begins and only proceeding where the numbers clearly stack up.
AI Governance
The policies, frameworks and controls that define how AI is used within an organisation, what it can access, what decisions it can make, how outputs are reviewed, and how it is managed over time. Governance is built into every Boxtree project to prevent ungoverned AI workarounds and protect the organisation.
Ungoverned AI Workarounds
Individual, unsanctioned use of AI tools by employees acting outside any organisational framework, typically using consumer AI products with company data in ways that create data security, compliance and consistency risks. Addressing ungoverned workarounds is a core objective of every Boxtree engagement.
AI Waste Identification
The process of identifying operational waste, unnecessary steps, waiting time, manual rework, duplicated effort, that is suitable for elimination through AI. Uses Lean's seven wastes framework as a lens, applied specifically to find where AI can intervene most effectively.
Embedded Implementation
An implementation approach in which AI solutions are built into existing workflows, trained into existing teams, and governed through existing management structures, rather than deployed as standalone tools. Embedded implementation maximises adoption and sustained value.
Artificial Intelligence Terms
Plain-English explanations of the AI terms you're most likely to encounter when exploring AI for your business operations.
Artificial Intelligence (AI)
Technology that enables computers to perform tasks that normally require human intelligence, understanding language, recognising patterns, making decisions and generating content. Most valuable in business where it handles high-volume, repetitive or analytical tasks faster and more accurately than humans can at scale.
Large Language Model (LLM)
An AI system trained on vast amounts of text that can understand, summarise, generate and translate language. LLMs power tools like ChatGPT and are the basis for AI applications that read documents, answer questions, generate reports and process unstructured text. Particularly useful for document analysis, communications and knowledge management.
Generative AI
AI systems that can create new content, text, images, code, data, based on patterns learned during training. In business operations, generative AI can draft documents, summarise data, generate reports and create structured outputs from unstructured inputs.
Machine Learning (ML)
A class of AI in which systems learn from data and improve over time without being explicitly reprogrammed. Particularly powerful where there is a large volume of historical data, enabling pattern recognition, anomaly detection and predictive analytics.
Natural Language Processing (NLP)
The branch of AI concerned with enabling computers to understand and work with human language, reading text, understanding meaning, extracting information. NLP powers document analysis, sentiment analysis, chatbots and many other business applications.
Process Automation
Using AI and software to automatically execute sequences of tasks that would otherwise require human intervention. Ranges from simple rule-based automation (RPA) to sophisticated AI-driven workflows that handle variability, exceptions and complex decision-making.
Robotic Process Automation (RPA)
Software that mimics human actions to automate repetitive, rule-based tasks, logging into systems, copying data, generating reports. Does not require AI but is often combined with it to handle exceptions. Well-suited to stable, high-volume administrative processes.
AI Agent
An AI system that can autonomously perform multi-step tasks, make decisions and interact with other systems to achieve a defined goal. AI agents can browse information, write and execute code, send communications and coordinate complex workflows, going well beyond answering questions.
Retrieval Augmented Generation (RAG)
A technique that enhances AI responses by allowing the system to pull in relevant information from your own documents and databases before generating an answer. Enables organisations to build AI tools that give accurate, up-to-date answers based on their own specific data, essential for knowledge management and compliance applications.
Prompt Engineering
The practice of crafting inputs to AI systems to reliably produce desired outputs. Well-designed prompts improve quality, consistency and usefulness of AI-generated content. In operational deployments, prompt engineering standardises how AI tools are used across an organisation.
Hallucination
When an AI system generates plausible-sounding but factually incorrect or fabricated information. One of the key risks of deploying AI in operational settings without appropriate governance. Boxtree addresses this through careful system design, output validation and governance frameworks that define when AI outputs should be reviewed before use.
API (Application Programming Interface)
A defined way for different software systems to communicate and share data. APIs are the plumbing that enables AI capabilities to be integrated into existing business systems, connecting AI solutions to your CRM, ERP, scheduling software or other platforms.
OCR (Optical Character Recognition)
Technology that converts images of text, scanned documents, PDFs, photographs, into machine-readable, searchable text. Foundational in document-processing workflows, enabling AI to work with information locked in paper or image-based documents. Widely used in invoice processing and contract management.
Data Pipeline
An automated process for collecting, transforming and moving data from one system to another. Often a necessary foundation for AI applications, ensuring the AI has access to clean, current, well-structured data. Many AI opportunities Boxtree identifies involve building data pipelines to connect systems that don't currently share information.
Fine-Tuning
Further training a pre-built AI model on organisation-specific data to improve performance for particular tasks. Fine-tuning enables a general-purpose AI to become expert in your specific domain, understanding your terminology, documents and way of working.
Computer Vision
AI's ability to interpret and understand visual information, images, video, documents, physical environments. Used in operational settings for quality inspection, document processing, inventory management and physical process monitoring.
Lean Methodology Terms
Lean is a proven operational improvement philosophy developed from the Toyota Production System. These are the core concepts that underpin Boxtree's diagnostic approach and our understanding of where AI creates the most value.
Lean Methodology
A systematic approach to operational improvement focused on maximising value for the customer while eliminating waste. Originally developed in Japanese manufacturing (Toyota), Lean has been applied successfully across manufacturing, services, healthcare, logistics and virtually every operational context.
Waste (Muda)
Any activity that consumes resources, time, money, people, without adding value. Lean identifies seven classic types, remembered by TIMWOOD: Transport, Inventory, Motion, Waiting, Overproduction, Overprocessing, and Defects. Identifying and eliminating waste is the starting point for every Boxtree engagement.
Value Stream Mapping (VSM)
A technique for visualising all steps, value-adding and non-value-adding, involved in delivering a product or service. VSM reveals where waste, delays, errors and inefficiencies occur, forming the basis for identifying improvement priorities. It is Boxtree's primary diagnostic tool during Discovery.
Gemba
Japanese for "the real place", the actual location where work happens. A core Lean principle is going to the Gemba to observe processes as they actually occur rather than as described in documentation. Boxtree consultants always spend time at the Gemba during Discovery, because real improvement starts with understanding real work.
Kaizen
The Japanese philosophy of continuous improvement through small, incremental changes, "change for the better." In Lean, Kaizen events are focused improvement workshops where a cross-functional team works intensively on a specific process to eliminate waste. The Kaizen mindset, that every process can always be improved, is foundational to Lean culture.
Standard Work
Documented best practice for completing a process consistently, the most efficient known method at a given point in time. Standard work reduces variation, enables training, and forms the baseline for further improvement. In AI implementation, standard work is critical: AI solutions must be built on well-understood, documented processes to function reliably.
Root Cause Analysis (RCA)
A structured method for identifying the underlying cause of a problem rather than treating its symptoms. The most common Lean tool for RCA is the "5 Whys", asking "why?" repeatedly until the true root cause is reached. Boxtree uses RCA to ensure AI solutions target root causes, not symptoms.
PDCA (Plan-Do-Check-Act)
An iterative four-step improvement cycle for testing and implementing changes. Plan: define the problem and solution. Do: implement on a small scale. Check: measure results. Act: standardise if successful or adjust and repeat. PDCA underpins Boxtree's phased implementation approach, build, test, measure, refine, then deploy.
Lead Time
The total elapsed time from a customer request to delivery of the result, encompassing all steps including waiting, queuing and delays. Reducing lead time is one of the most visible and impactful outcomes of Lean and AI improvement, with direct benefits for customer experience and capacity.
Cycle Time
The time required to complete one unit of work, one transaction, one document, one product. Distinct from lead time: lead time includes all waiting and delays, cycle time measures only active processing. Reducing cycle time through AI automation directly increases throughput and capacity.
Takt Time
The rate at which work must be completed to meet customer demand, calculated by dividing available working time by the number of units demanded. Takt time is the heartbeat of a Lean process; understanding it enables resources to be matched to demand and over/under-production to be identified and addressed.
Kanban
A visual system for managing workflow, originally using physical cards to signal when work should move to the next stage. Kanban limits work-in-progress, makes bottlenecks visible and ensures smooth flow. Digital kanban systems are widely used in knowledge work and integrate well with AI-driven workflow management.
5S
A workplace organisation methodology: Sort, Set in order, Shine, Standardise, Sustain. Creates an organised, visual workplace that reduces waste and supports standard work. In AI deployment, 5S principles ensure systems, data and processes are well-maintained and easy to improve over time.
Want to learn more?
Talk to us about how Lean AI methodology applies to your specific operations and challenges.
Get in touch →