AI buzzwords explained for lawyers 

1. Artificial Intelligence (AI) 

In plain English:  

Software that can perform tasks which would normally need human thinking – like reading a document, spotting patterns, or making a decision. 

When a legal software provider says a product is “AI-powered,” it usually means the software can do something intelligent on its own, such as categorising emails, flagging risks in contracts, or suggesting next steps in a case. It does not mean there is a robot sitting in a server room. AI is a broad umbrella term, and most of the words below sit underneath it. 

2. Machine Learning (ML) 

In plain English:  

A type of AI that learns from examples rather than following a fixed set of rules. 

Think of it like training a junior solicitor. Instead of writing them an instruction manual for every scenario, you show them hundreds of examples, and they start to recognise the patterns themselves. Machine learning works the same way – feed it enough data, and it figures out how to make predictions or decisions. It is the technology behind things like spam filters and, increasingly, legal document review tools. 

3. Large Language Model (LLM) 

In plain English:  

An advanced AI that has been trained on vast amounts of text so it can understand and generate human language. 

ChatGPT is the most well-known example. LLMs can summarise lengthy documents, draft correspondence, answer questions in natural language, and much more. For law firms, the practical benefit is significant: an LLM can read a 50-page lease and pull out the key commercial terms in seconds. The “large” part refers to the sheer scale of training data – we are talking about billions of pages of text.

4. Generative AI (GenAI) 

In plain English:  

AI that creates new content – text, images, summaries, even code – rather than just analysing what already exists. 

This is the category of AI that has captured public attention since late 2022. For legal professionals, generative AI is most useful for drafting, summarising, and rephrasing. Ask it to draft a first-pass client update letter based on case notes, and it will produce something usable in seconds. It generates; it does not just search and retrieve. 

5. Agentic AI 

In plain English:  

AI that can independently plan and carry out multi-step tasks, making decisions along the way rather than waiting for instructions at every stage. 

Standard AI tools respond to a single prompt and give you an answer. Agentic AI goes further – you give it a goal, and it works out the steps to get there. In a legal setting, this could mean an AI that receives a new instruction, identifies the relevant documents, drafts the initial correspondence, populates the key dates in your case management system, and flags anything that needs human review – all without you having to oversee each step individually. It is still early days, but agentic AI is the direction travel for legal technology, and it is why you will hear the term increasingly in vendor conversations. The important question to ask is: what guardrails are in place to keep a human in the loop when it matters? 

6. Natural Language Processing (NLP) 

In plain English:  

Technology that helps computers understand, interpret, and respond to everyday human language. 

NLP is the reason you can type a question into legal research software in plain English – rather than needing to use complex search strings – and still get relevant results. It powers everything from voice assistants to contract analysis tools. When your case management system lets you search by describing what you are looking for in your own words, NLP is doing the heavy lifting behind the scenes. 

7. Algorithm 

In plain English:  

A set of step-by-step instructions that tells a computer how to solve a problem or complete a task. 

This one gets thrown around a lot, often to make things sound more technical than they are. A recipe is an algorithm. So is the process your firm follows for onboarding a new client. In the AI world, algorithms are simply the rules and calculations that drive how the software makes decisions. When someone says “our algorithm detects risk,” they mean the software follows a defined process to flag potential issues. 

8. Neural Network 

In plain English:  

A computing system loosely inspired by the human brain, designed to recognise patterns in data. 

Neural networks are made up of layers of interconnected nodes (think of them as simplified brain cells). Data passes through these layers, and each one refines the output a little further. They are particularly good at tasks where the rules are hard to write down explicitly – like recognising handwriting on scanned documents, or understanding the sentiment of a witness statement. 

9. Deep Learning 

In plain English:  

A more advanced form of machine learning that uses neural networks with many layers to tackle complex problems. 

The “deep” refers to the number of layers in the neural network, not the depth of understanding (though the results can be impressive). Deep learning is what makes modern AI capable of things like accurate speech recognition, image analysis, and sophisticated document understanding. If standard machine learning is a junior solicitor spotting patterns, deep learning is a senior associate who can handle far more nuanced work. 

10. Training Data 

In plain English:  

The information used to teach an AI system how to do its job. 

An AI is only as good as the data it learns from. If you train a contract review tool using thousands of well-drafted commercial leases, it will get very good at spotting unusual clauses in similar documents. If you train it on irrelevant data, the results will be poor. This is why legal-specific AI tools tend to outperform general-purpose ones – they have been trained on legal documents, not Wikipedia articles.

11. Prompt 

In plain English:  

The instruction or question you give to an AI tool to tell it what you want. 

When you type a question into ChatGPT or a similar tool, that is a prompt. The quality of the output depends heavily on how well you frame the input. Asking “Summarise this contract” will give you a very different result to “Summarise the key obligations, termination provisions, and liability caps in this contract.” Getting good at writing prompts – sometimes called prompt engineering – is becoming a genuinely useful professional skill. 

12. Hallucination 

In plain English:  

When an AI confidently generates information that is incorrect or entirely made up. 

This is arguably the most important term on this list for lawyers. LLMs do not “know” things the way humans do – they predict what word should come next based on patterns. Sometimes that leads to plausible-sounding but completely fictitious output, including invented case citations. Several lawyers globally have already been sanctioned for submitting AI-generated briefs containing fabricated cases. The lesson: always verify AI output, especially anything that will end up before a court. 

13. Retrieval-Augmented Generation (RAG) 

In plain English:  

A technique that makes AI more accurate by feeding it specific, relevant information to reference before it generates a response, rather than relying solely on what it learned during training. 

RAG is one of the most important developments for legal AI because it directly tackles the hallucination problem. Instead of asking an LLM to answer a question purely from memory, a RAG system first searches your firm’s own documents, knowledge base, or case files, retrieves the relevant information, and then passes it to the AI along with the question. The result is an answer grounded in real, verifiable sources rather than the model’s best guess. When a legal software vendor tells you their AI “only draws from your firm’s data,” they are almost certainly describing a RAG-based approach. It is the difference between an AI that invents plausible-sounding answers and one that can point you to the specific document it relied on. 

14. Automation vs. AI 

In plain English:  

Automation follows fixed rules (“if this, then that”). AI can make judgments and handle situations it has not been explicitly programmed for. 

This distinction matters because not everything branded as AI is actually AI. Automatically sending a standard client care letter when a new matter opens is automation. Reading an incoming email and routing it to the right fee earner based on its content is AI. Both are valuable, but they are fundamentally different. Many of the biggest efficiency gains in law firms actually come from smart automation, not AI. 

15. Optical Character Recognition (OCR) 

In plain English:  

Technology that converts scanned documents, PDFs, or images of text into actual editable, searchable text. 

If your firm deals with paper files, bundles of scanned documents, or image-based PDFs, OCR is what makes that content usable digitally. Modern OCR combined with AI can handle handwriting, poor-quality scans, and unusual layouts far more reliably than older tools. It is a foundational technology for digitising legal archives and making historical case files searchable. 

16. Predictive Analytics 

In plain English:  

Using historical data and AI to forecast what is likely to happen next. 

In a legal context, this could mean predicting the likely outcome of a case based on similar matters, forecasting how long a transaction will take to complete, or identifying which clients are at risk of leaving. It is not crystal-ball gazing – it is pattern recognition applied to your firm’s own data. The more quality data you have, the more accurate the predictions become. 

17. Chatbot vs. AI Assistant 

In plain English:  

A chatbot follows a script. An AI assistant understands context and can handle open-ended questions. 

Early chatbots were essentially decision trees – click option A, get response B. Modern AI assistants are far more flexible. They can understand a question phrased in multiple ways, remember the context of a conversation, and provide nuanced responses. In legal software, this shows up as intelligent help features that can answer practice-specific questions rather than just pointing you to a generic FAQ. 

18. Bias (in AI) 

In plain English:  

When an AI system produces unfair or skewed results because of problems in its training data or design. 

AI learns from historical data, and historical data reflects historical decisions – including unfair ones. If a recruitment AI is trained on a decade of hiring data from a firm that predominantly hired from a narrow set of universities, it may unfairly penalise candidates from elsewhere. For law firms, understanding AI bias is important both as a risk management issue and as a matter of professional ethics, particularly in areas like criminal justice and employment law. 

19. Responsible AI / Ethical AI 

In plain English:  

The practice of developing and using AI in a way that is fair, transparent, and accountable. 

As AI becomes more embedded in legal workflows, questions of responsibility become critical. Who is accountable when an AI tool gives bad advice? How do you ensure client data is handled properly? Responsible AI is not just a buzzword – it is a framework for making sure that technology is deployed thoughtfully. The Solicitors Regulation Authority and Bar Standards Board are both paying increasing attention to how legal professionals use AI tools. 

20. Workflow Automation 

In plain English:  

Using technology to handle repetitive, rules-based tasks automatically, so people can focus on higher-value work. 

This is where AI and automation meet the day-to-day reality of running a law firm. Automatically generating standard documents, chasing overdue invoices, updating case milestones, routing incoming enquiries – these are all examples of workflow automation. The best legal software builds this in natively, so it works within the tools your team already uses rather than requiring a separate system. 

Understanding the terminology is a useful starting point, but the real question for most legal professionals is a practical one: where does this technology actually show up in my day-to-day work, and how does it make a difference? 

The answer, increasingly, is your case management system. Modern case management software is no longer just a digital filing cabinet - it is becoming the central hub where many of these AI capabilities come together in ways that directly improve how legal work gets done. 

Smarter document handling 

Several of the terms above – OCR, NLP, and large language models – converge in how case management platforms handle documents. Rather than manually reading, tagging, and filing every incoming document, AI-enabled systems can extract text from scanned files (OCR), understand what type of document it is (NLP), and even summarise its contents or flag key clauses (LLMs). For firms handling high volumes of conveyancing packs, bundles, or discovery documents, this is not a marginal improvement; it fundamentally changes the speed and accuracy of document workflows. 

RAG plays a critical role here. When a fee earner asks the system a question: “What are the break clauses in the Smith lease?”, a RAG-enabled case management system retrieves the relevant documents from your firm’s own files, passes them to the AI, and generates an answer grounded in actual case data rather than generic training. This is how modern legal AI delivers trustworthy, source-backed responses instead of the kind of confident-sounding guesswork that has landed other professionals in trouble. 

Workflow intelligence 

Workflow automation and AI work hand in hand within a case management system. The automation handles the predictable, rules-based tasks; generating standard letters, triggering reminders, and updating case milestones. AI adds a layer of intelligence on top: routing incoming enquiries to the right team based on the content of the message, suggesting next steps based on similar matters, or flagging a case that is at risk of missing a deadline before anyone has noticed. The combination means fewer things slip through the cracks, and fee earners spend more time on work that actually requires legal judgment. 

This is also where agentic AI is beginning to make its presence felt. Rather than automating one task at a time, agentic systems can orchestrate entire sequences: receiving a new instruction, identifying and assembling the relevant documents, pre-populating key dates, and drafting initial correspondence, all while keeping a human in the loop for approval at critical stages. It is the next evolution of workflow intelligence, and the firms that benefit most will be those whose case management systems are built to support it natively. 

Turning case data into actionable insight 

Every matter your firm handles generates data – timelines, outcomes, costs, client interactions. Predictive analytics and machine learning can turn that historical information into forward-looking intelligence. How long is this type of matter likely to take? Which matters are most profitable? Where are bottlenecks forming? A case management system that captures this data consistently gives AI something meaningful to work with. Without that structured foundation, even the most sophisticated AI tools have nothing to analyse. 

Improving the client experience 

AI assistants, NLP-powered search, and generative AI all contribute to a better experience for clients. Imagine a client portal where enquiries are understood and triaged instantly, where status updates are generated automatically based on live case data, and where routine questions are answered without a fee earner needing to pick up the phone. None of that requires futuristic technology – it requires a case management platform that has been designed with these capabilities built in, not bolted on as an afterthought. 

The importance of getting it right 

Of course, with all of this capability comes responsibility. The risks around AI hallucination, bias, and data security are real, and the legal profession rightly holds itself to a higher standard than most. Techniques like RAG help reduce hallucination risk by grounding AI responses in verified, firm-specific data, but they do not eliminate the need for human oversight entirely. That is why the choice of technology partner matters as much as the choice of technology. You need a provider that understands the regulatory environment, takes data governance seriously, and builds AI features that are transparent and auditable rather than opaque. 

Adopting AI does not mean ripping up what works. The firms getting the most value are the ones choosing technology that embeds intelligence into the workflows they already rely on. 

At Access Legal, that is exactly how we approach it. Our case management software, CaseMatters Evo, is built with AI and automation at its core, designed specifically for the way legal professionals work. From intelligent document handling and workflow automation to predictive insights and smarter client communication, the capabilities described in this guide are not abstract concepts; they are features your team can use today. 

If you are ready to see how AI-powered case management can make a practical difference in your firm, take a look at our virtual product tour.