AI assistant for a law firm
It works through an incoming enquiry, establishes the subject and collects the dates, amounts and documents. The lawyer gets a prepared file rather than a line saying somebody wants a consultation.
- Buildfrom $1,700
- Timeline2–4 weeks
- Supportfrom $570/mo
Who this suits: Practices where the first conversation is held by a lawyer billing by the hour, and a good share of enquiries are not in their area at all.
What is in the first version
- intake questionnaire
- document upload
- conflict check
- case file
The figures above are the package this sits on: AI agent or process automation. Anything past its edges the calculator adds before the work starts, not after.
How the path goes
- 01
The enquiry starts as an interview
The assistant first establishes the type of matter, and the questions that follow belong to that type.
- 02
Facts come one at a time
It asks for what is missing, accepts photographs, and marks explicitly whatever the file still does not contain.
- 03
The lawyer opens a file
The subject, the timeline, the documents and the question are on the desk, so the conversation starts at the substance.
What decides whether this works or annoys people
It collects facts and assesses nothing
No prognosis, no limitation period, no value of a claim, even approximately: that wording is legal advice.
Your opponent may be your client
The parties are checked against your records before the first substantive question, because a late conflict still means declining.
Confidentiality starts before the engagement does
Where the data sits, how long it lives and what reaches a model is settled before launch, in the contract.
What to measure once it is live
- enquiries filtered out before the lawyer
- completeness of the file at intake
- time to the first reply
- enquiries that reach an engagement
Why this comes out faster
The shape of this one is known: the states, the edge cases and the things that usually go wrong have been decided before. Nothing here is a template, and the saving is not in your half of the work. It goes into your process, your content and the systems this has to talk to, which is the part nobody can have solved in advance.
A likely stack for this
Picked against the task when we scope it, not decided in advance. This is the shelf it usually comes off.
- Assistant and serverPythonFastAPIPostgreSQLRedisTelegram Bot API
- Models and documentsOpenAIAnthropicpgvectorS3
- Running itDockern8nGrafana
Price it yourself, right here
Five steps, and you can see the number without leaving a contact. The estimate accounts for the kind of work, what you already have and what it will need inside.
These bills do not come from us
- AI model calls at the provider’s rate
- Paid access to registries and court records
- Hosting and storage for scans and threads
- Telephony and SMS for calls and reminders
Asked before the first call
- Is this legal advice given without a lawyer?
- No, and that is a limit inside the system rather than a line in a prompt. The list of what it never says, including prospects and limitation periods, you approve before launch.
- Who answers for it if the assistant misunderstands somebody?
- The lawyer does, which is why the assistant decides nothing. Every question and answer sits in a log beside the file, in the enquirer’s own wording.
- Can it feed into the case system we already run?
- Yes, the file goes where your matters already live. Without such a system the first version keeps its own list, with export to standard formats from the start.