ChatGPT is a remarkable general-purpose tool. Lawyers use it to outline arguments, tighten prose and think through problems. But a general-purpose chatbot and an AI legal assistant built for law firm work solve different problems. The difference isn’t intelligence. It’s what the tool knows before you ask your first question.
This article compares the two on the five things that matter most in daily legal work: matter context, sources, confidentiality, drafting and workflow.
The briefing problem
Every useful answer about a legal matter depends on the facts of that matter. A general chatbot starts every conversation knowing none of them.
So before it can help, someone has to brief it. Find the right documents. Decide which ones are safe to share. Copy in the relevant parts. Explain who the parties are and what happened when. Then ask the question.
For a quick question, that’s fine. For real casework, the briefing becomes the job. By the time the chatbot understands the matter, a paralegal could have answered the question directly.
An AI legal assistant built for law firms flips that sequence. It has already read the matter: the documents, the emails, the intake answers. The question is the only input it needs.
Comparison at a glance
What it knows. A general-purpose chatbot knows what you paste into the conversation. An AI legal assistant built for law firms knows the whole matter file, kept current.
Sources. A chatbot often gives none, or ones that are hard to verify. A legal assistant cites the document each answer came from.
Client data. A chatbot is governed by its plan and settings. A legal assistant is governed by a vendor agreement built for client files.
Drafting. A chatbot writes in a general format. A legal assistant drafts from the firm’s own templates.
Where it lives. A chatbot sits in a separate window. A legal assistant sits inside the firm’s matter workflow.
1. Matter context
Legal questions are rarely general. “What did the client say about the second account?” can only be answered by someone who has read the intake answers, the emails and the statements.
A general chatbot can reason well about whatever text you give it. It can’t know what you didn’t give it. That creates a quiet risk: the answer is only as complete as the excerpt you chose.
A legal assistant working from the full file can surface the email you forgot existed, or the statement that contradicts the intake answer.
2. Sources and verification
Legal work runs on evidence. An answer is only useful if your team can check it quickly.
General chatbots are known to produce confident answers that turn out to be wrong, and courts have sanctioned lawyers who filed AI-generated citations without checking them. The lesson isn’t to avoid AI. It’s to use AI that shows its work.
When every answer links to the document, page or message it came from, verification takes seconds. That changes AI from something you have to double-check from scratch into something you can supervise.
3. Confidentiality
The ABA’s Formal Opinion 512, issued in July 2024, applies the existing duties of competence, confidentiality, communication and supervision to generative AI. Among other things, it asks lawyers to understand how a tool handles client information before putting that information into it.
For a general chatbot, the answer depends on which plan you use and how it’s configured. Read the terms carefully, including whether conversations can be used to improve the provider’s models.
For a legal assistant, ask the vendor directly:
Is the platform independently audited?
Is client data ever used to train models?
Who inside the firm, and outside it, can access each matter?
Deliberately.ai is SOC 2 compliant and never uses client data for model training. Whatever tool you choose, get those answers in writing.
4. Drafting
A chatbot can draft a decent generic letter. But a law firm doesn’t send generic letters. It sends letters in its own format, with its own signature block, built on the actual facts of the matter.
An AI legal assistant that drafts from the firm’s own Word templates and the matter file produces a first draft that already looks like your work. The edit is about judgment, not reformatting.
5. Workflow
Every new tool competes for attention. A chatbot in a separate browser tab means switching context, copying text out and pasting results back.
A legal assistant that lives where the matter lives, and that you can reach by email from a courthouse hallway, gets used. Tools that get used are the only ones that save time.
When ChatGPT is the right tool
This isn’t an argument against general AI. It’s excellent for:
Brainstorming arguments and counterarguments on a hypothetical
Explaining an unfamiliar concept in plain language
Editing your own non-confidential writing for clarity
Drafting internal templates, checklists and training material
The rule of thumb: if the task depends on general knowledge, a general tool works. If it depends on the facts of a client’s matter, use a tool that already has them.
Questions to ask before choosing
Does it know our matters, or do we have to brief it each time?
Does every answer cite its source?
Is client data used for training, and is the platform audited?
Can it draft from our templates?
Will our team actually use it, every day?
The bottom line
ChatGPT is a brilliant generalist. An AI legal assistant is a specialist that has already read the file. Most firms will use both. The mistake is asking the generalist to do the specialist’s job and spending the saved time on the briefing instead.
If you’d like to see the difference on one of your own matters, read about Del, the AI legal assistant that already knows the case, or start a free trial.
Related reading: ChatGPT prompts for lawyers and law firms
