Chapter 0 · Introduction
00 Introduction 3 lessons
01 How can AI help my practice and day-to-day work? 5 lessons
02 Can I trust it? 5 lessons
03 Am I allowed to use it? 7 lessons
04 How do I choose the best tools? 6 lessons
05 How do we make it work across the firm or legal team? 7 lessons
06 How do I get better results? 3 lessons
07 What does this mean for my career and my firm? 4 lessons
08 Where and how do I start now? 4 lessons
Lesson 0.3 Introduction 8 min read
Survival Glossary (in Plain Words)
The 21 words you'll meet in this guide, in vendor demos and in your IT team's emails. Each gets a short definition, a concrete example and a small picture. Skim them now and come back when a word trips you up.
The model
LLM / model
A large language model is a computer program that has read an enormous amount of text and learned to guess which word should come next. Doing that guess over and over is how it writes answers. GPT, Claude, Gemini and Mistral are families of these models.
Example When I type 'The tenant shall give written…', the model continues with 'notice' because that word usually follows those words in everything it has read. It didn't look at any lease.
The tenant shall give written…
- notice 64%
- consent 21%
- reasons 9%
Token
The small pieces of text a model actually reads and writes: a short word, or a chunk of a longer one. Limits and prices are counted in tokens. In English, 100 tokens is roughly 75 words.
Example A 40-page lease is about 20,000 words, which is about 27,000 tokens. That figure is what counts against the model's limit, and what the bill is based on.
Indemnification clause
Indemnificationclause
2 words · 4 tokens
Prompt
Everything you send the model in one go: your question, your instructions and any document you attach. The clearer the prompt, the better the answer.
Example 'Summarize this' got me a vague page. 'You act for the tenant. Summarize this lease in five points for a client who isn't a lawyer' got me exactly what I needed, from the same document.
Context window
The model's working memory: how much text it can hold in mind at once, counting your instructions, any documents and the conversation so far. Anything beyond that limit is cut off or forgotten.
Example I pasted a 300-page textbook into the chatbot and asked about chapter 2. It answered from the last few chapters only: the beginning had been pushed out of its context window.
- Instructions
- Your documents
- Conversation so far
Full? Older material gets cut first.
Chatbot
A model you talk to in a chat window, like ChatGPT, Claude or Le Chat. You type, it replies, and each new reply takes the earlier messages into account.
Example I asked Le Chat whether a landlord can refuse to renew. It asked which country. I said France, and from then on its answers assumed French law without my having to repeat it.
Can the landlord refuse to renew?
It depends on the lease and the governing law. Which country?
France.
Hallucination
When a model states something false or made up, such as a court case that never existed, in the same confident tone it uses for things that are true. It isn't lying; it is guessing, and it doesn't flag the difference.
Example In Mata v. Avianca (2023), two New York lawyers filed a brief citing six cases ChatGPT had invented, complete with realistic-looking citations. The court fined them.
Harlow v. Meridian Holdings
758 F.3d 412 (2d Cir. 2014)
- Right format
- Real court and reporter
- No such case
What's built around it
RAG
Retrieval-augmented generation. Before answering, the system looks up the relevant passages in a chosen set of documents and hands them to the model, so the answer rests on those passages and can cite them, rather than on memory alone.
Example I asked 'When can we terminate?'. The tool first found §12 of the lease and the side letter in our files, then answered citing both, so I could check every point against the source.
- 1“When can we terminate?”
- 2 Lease §12Side letter
- 3Answer based on them, citing [1] [2]
Skill
A set of written instructions, saved once and reused, that teaches an assistant how to do one task the way you want it done. Think of a recipe card, sometimes with templates or small scripts attached.
Example I wrote our NDA review checklist into a skill once. Now whenever anyone asks the assistant to review an NDA, it checks term and scope and flags unusual clauses, exactly the way we do.
nda-review/SKILL.md
## When to use
Reviewing an NDA for a client
## Steps
1. Check term and scope
2. Flag unusual clauses
MCP
Model Context Protocol: a standard plug that lets an assistant connect to outside tools and data, such as your cloud storage, a case-law database or an email account. One standard, so each tool doesn't need its own custom wiring.
Example Through MCP connections, the assistant searched our cloud storage and a case-law database from the same chat, with no copy-pasting between them.
Plugin
A bundle that adds several things to an assistant in one install: skills, MCP connections and ready-made commands. It is how one person's setup gets shared with a whole team.
Example Installing the contracts plugin gave everyone on the team the same three skills, the connection to our document system and two commands, in a single step.
contracts-plugin
3 skills 1 MCP connection 2 commands
Agent
A model set up to work toward a goal on its own, step by step: it decides what to do next, uses tools, checks the result and keeps going until the job is done or it gets stuck. Less like asking a question, more like delegating a task.
Example I gave it access to the data room and asked for a table of change-of-control clauses. It opened all 40 contracts one by one, pulled out the clauses, built the table and handed it back for me to review.
- Open the data room
- Read 40 contracts
- Extract key clauses
- Build the summary table
- Hand back for your review
Harness
Everything built around a model to turn it into a usable product: its standing instructions, the files and tools it can reach, and its safety checks. Same model, different harness, different results.
Example The same model gave a generic answer in a public chatbot and a sourced answer in our firm's house style inside our legal tool. The model hadn't changed; the harness had.
Harness
Model
API
An application programming interface: the way one program talks to another, with no human at a keyboard. Companies use a model's API to build it into their own software, usually paying per token.
Example Our document system sends each new contract to a model's API with the instruction 'Summarize §8' and files the reply automatically. No chat window involved.
POST /v1/messages
{ "prompt": "Summarize §8" }
200 OK
{ "text": "§8 caps…" }
Billed per token, in and out.
Benchmark
A standard test used to score and compare models, a bit like an exam league table. Useful for a rough ranking, but a high score doesn't prove the model will do good work on your matters.
Example A model can top a public leaderboard and still miss the break clauses in your leases. So we tested the shortlist on 20 of our own files before choosing.
- Model A 86%
- Model B 79%
- Model C 64%
Your contracts aren't the test.
Where it runs
Open-source
Software whose inner workings are published for anyone to inspect, use and change. An open-weight model is one you can download and run on your own computers, instead of only through the maker's service.
Example Llama and many Mistral models can be downloaded and run on your own servers. Claude and Gemini can only be used through their makers' apps and APIs.
Open: download it, inspect it, run it on your own servers
Closed: use it only through the vendor's app or API
Git
A system that keeps every version of a set of files and records who changed what, when and why. Any earlier version can be brought back. Like track changes, but for whole folders and forever.
Example The client asked why clause 8 had changed. Git showed who rewrote it, on which Monday and with what note, and we restored Friday's draft in one step.
- a1f3c9 Client edits Today
- 7be20d Clause 8 redone Mon
- 3c94e1 First draft Fri
Markdown
A simple way to mark up plain text with symbols: # for a heading, ** for bold, - for a bullet point. AI tools read and write it naturally.
Example Those stray ## and ** that appear when you paste an assistant's answer into Word are Markdown marks that weren't converted into real formatting.
## Notice
**30 days**, in writing
- by registered mail
Notice
30 days, in writing
• by registered mail
Server
A computer that runs around the clock and answers requests from other computers over a network. When you use an online AI tool, your text leaves your device and is processed on someone else's server.
Example When I paste a client email into an online chatbot, it leaves my laptop and is processed on the provider's server, possibly in another country.
Wherever the provider runs it
Database
An organized store of records that software can search, sort and update in an instant: a filing cabinet that files itself. A list of matters, clients and deadlines, for example.
Example Our matters live in a database, so 'every November deadline for Dupont SA' comes back in a second, instead of a scroll through a spreadsheet.
| Matter | Client | Due |
|---|---|---|
| Lease renewal | Dupont SA | 14 Nov |
| NDA review | Acme Ltd | 18 Nov |
| Appeal | M. Leroy | 2 Dec |
Cloud provider
A company that rents out computing power and storage in its own data centers, such as AWS, Microsoft Azure, Google Cloud or OVHcloud. Most online tools run on one of them, so where that provider stores your data matters.
Example Our legal AI vendor runs on Microsoft Azure, so we asked which region. Paris keeps client data in the EU; Virginia would not.
AWSMicrosoft AzureGoogle CloudOVHcloud
Region:Paris or Virginia?
On-prem
Short for on-premises: software that runs on computers your organization owns and controls, in your own building, instead of in a provider's cloud. More control, more upkeep.
Example The firm runs an open-weight model on its own servers, so client files never leave the building. The trade-off: its IT team now handles every update and security patch itself.
Your firm
Your servers Your team
Updates and security: yours too.