Scoped to your work
Answers come from this matter's documents and your playbook only. The assistant never reaches outside the work in front of it.
AI assistant
Ask any contract, matter, or case a question and get an answer grounded in its documents and your playbook. It cites its sources, reads even scanned files, and never reaches outside the work in front of it.
Assistant
Scoped to: Vantage Spectrum Licence
The problem
Most AI tools answer from everything and cite nothing. For legal work, that is the wrong trade. You need to know where an answer came from, and you need to know it stayed inside the matter.
What it does
Grounded in the documents of the contract, matter, or case in front of you.
Every answer cites the source it came from, so you can check it.
Reads even scanned files, so nothing is locked out of a question.
Scoped to one piece of work. A conversation about one matter never sees another.
How it works
The assistant is built around one rule: an answer is only as good as the source behind it.
A contract, a matter, or a case. The question goes to that work's documents and your playbook, nothing else.
Every answer cites the documents it came from, clause by clause. Scanned files included.
Open the cited source next to the answer and read it. The claim is only as strong as its source, so the source is always in reach.
What was asked and what was answered is kept with the matter, so the work can be replayed or audited.
Assistant
Scoped to: Harbor Point Acquisition
Three of the eleven agreements require consent. The Meridian supply agreement requires written consent before completion (cl. 14.2). Kestrel's services agreement treats a change of control as an assignment needing approval (cl. 9.1). The Orbus licence, a scanned document, carries a termination right on change of control (cl. 17.4).
Answers come from this matter's documents and your playbook. Nothing outside it.
In the product
Ask a contract, a matter, or a case. The answer comes back grounded in that work's documents and your playbook, with the sources it used.
Answers come from this matter's documents and your playbook only. The assistant never reaches outside the work in front of it.
Every answer points back to where it came from, so you can open the document and check it yourself.
Text recognition turns paper and scans into documents the assistant can search and cite. Nothing is locked out because it arrived on paper.
Several angles of analysis on a document, each finding traceable to the source, written the way a careful lawyer would.
A document in one matter is never visible to an assistant in another. Cross-matter confidentiality is built in.
What was asked and what was answered, kept with the matter, so the work can be replayed or audited.
A conversation about one case does not see another. A matter does not leak into another matter. The heavy AI work runs on infrastructure built for it and never trains on your data.
In practice
Not party tricks. The reads and lookups that eat a working week.
Which agreements carry change-of-control consent? Asked once, answered with citations across every uploaded agreement.
A new lawyer on the matter asks the file what has happened so far, and reads the answer with the sources open.
Find the indemnity language across a matter's documents without opening them one by one.
Ask the review why it flagged a finding, and follow the citation back to the page it came from.
Who it's for
For the lawyer
A fast answer you can actually check, drawn from the file in front of you, with the source cited.
For the function
A way to put AI to work that respects confidentiality, stays in scope, and leaves a record.
Bring a contract or a case. We will show you the assistant answering from your own documents.