Glossary
Vibe Coding — Definition for Lawyers
Vibe coding means building software by describing what you want to an AI tool in plain language and iterating on the result. What lawyers build with it, and its limits.
On this page
Vibe coding is building software by describing what you want to an AI model in plain language, running what it produces and asking for changes until it works, without writing or necessarily reading the code yourself. For lawyers it turns a spreadsheet-sized irritation into a working tool in an afternoon. It does not turn a lawyer into a software engineer, and the difference matters as soon as client data is involved.
How it works
The loop is short. You describe the tool, the AI generates the code and usually a live preview, you try it, and you describe what is wrong. Repeat. The tools range from chat assistants that produce a self-contained web page, through browser-based app builders that also host the result, to coding agents such as Claude Code that work on files on your own machine.
What makes it work for lawyers is that the hard part of a small legal tool is rarely the code. It is knowing what the rules are, which edge cases matter and what the output must look like. Lawyers have that knowledge; the AI supplies the syntax.
What lawyers build, and what they should not
| Good fit | Poor fit |
|---|---|
| A limitation-period or fee calculator for one jurisdiction | A firm-wide docketing system |
| A client intake questionnaire that routes to the right team | Anything that gives legal advice to clients without review |
| A decision tree for a regulatory threshold test | A system of record for client money or conflicts |
| A tool that sorts signature pages or renames files in bulk | Software others depend on that only one person can maintain |
| A clause comparison or side-letter tabulator for one deal | Integrations with the DMS or billing system built without IT |
The pattern on the left is a bounded tool that does one thing, used by people who can tell when its output is wrong. The decision-tree playbook and the client-intake playbook each walk through one build from blank screen to working app.
A specification that gets better results
The most common failure is prompting with a one-line idea. Write the spec as a user experience first:
Build a single-page web app for internal use at a law firm.
User: an associate checking whether a share transfer needs notification.
Inputs: target's country, acquirer's country, percentage acquired,
sector (dropdown: defence, energy, telecoms, media, other).
Logic: [paste your decision rules as numbered if/then statements]
Output: "Notification likely required", "Not required on these facts" or
"Refer to partner", with the rule that triggered the result quoted.
Constraints: no data leaves the browser; no external calls; no login.
Include five test cases with expected outputs and a button to run them.
The test cases are the part lawyers skip and should not. They are how you know the tool applies your rules rather than plausible-looking ones.
Limits and data-handling cautions
Three risks recur. Data handling: an app builder may host your app and store what users type, so prototype on synthetic data and check where anything is stored before real information goes near it. Correctness: the code can look right and implement the rule wrongly, so verify against cases you know. Maintenance: a tool nobody else understands becomes a liability when its author moves on. The build-or-buy guide sets out when a vibe-coded tool should be replaced by a supported product, and the agent security guide covers where coding agents stop being safe to run on firm machines.
Where this leaves you
Pick one repeated task that annoys you, write the specification and test cases on paper, and build it with synthetic data. If it survives a week of real use by colleagues, decide with IT whether it stays. Module 5 of the AI Lab for Lawyers course covers vibe coding calculators, questionnaires, decision trees and client-intake apps without code.
Frequently asked questions
Do I need to know how to program to vibe code?
No, and that is the point of the term. You describe the tool, the AI writes the code, and you judge the result by using it. What you do need is the ability to specify precisely what the tool should do, to test it against cases where you know the right answer, and to recognise when a problem is beyond what you can safely maintain.
Can I use a vibe-coded app with client data?
Not without the same scrutiny you would give any software vendor. A prototype built in an afternoon may store data in places you did not choose, log inputs, or rely on a hosting service whose terms you have not read. Build and test with synthetic data, then have whoever is responsible for IT and data protection review the app before real client information goes near it.
Put this into practice
AI Lab for Lawyers: a live, hands-on course for practising lawyers
4 live sessions (8 hours), recordings included, taught by Dr. Niklas Schmidt of Wolf Theiss. Rated 4.7/5 from 381 ratings; alumni from Norton Rose Fulbright, Paul Hastings, Pillsbury and Taylor Wessing.
Related reading
Playbooks
Vibe-Code a Client Intake App as a Lawyer (Lovable, Claude Artifacts, v0)
Vibe-code a client intake app as a lawyer with Lovable, Claude Artifacts or v0: the spec, four exact prompts, a fee-band calculator and the IT handover.
8 min read
Playbooks
Build a Legal Decision-Tree App for Clients Without Code
Build a legal decision-tree app for clients without code in a day: write the logic as a node table, generate it with Claude or Lovable, and test the edge cases.
11 min read
Tools
Claude Code for Lawyers: Should Non-Programmers Use a Coding Agent?
Claude Code is a terminal coding agent. When lawyers should use it for bulk document work and small tools, two worked examples with guardrails, safer routes.
9 min read
Guides
Build or Buy Legal AI? Latham's GPUs, Harvey's Seats and the Firms in Between
Build vs buy legal AI in 2026: rent Harvey or Legora, assemble on Claude or Astra with MCP, or build like Latham on own GPUs. Costs, control, decision tree.
11 min read
Guides
Agent Security for Lawyers: Prompt Injection, Data Leaks and Guardrails
AI agent security for law firms: prompt injection via hostile documents, the Grok Build and Cowork incidents, and the allowlists and gates that contain them.
11 min read
Guides
Junior Lawyers and AI: Training, Hiring and the Missing Rung
Junior lawyers and AI in 2026: Big Law hired 538 fewer graduates, 2% of UK juniors say AI helps them learn, and firms are rebuilding training with AI seats.
11 min read