ReframeWhat law firm enterprise search has to do
A law firm keeps decades of agreements, opinions and closing sets in iManage. A partner asks which side letters granted a fee discount, or what the firm accepted from one counterparty last year. A keyword search returns documents that contain the words, and the partner still reads every document.
| Requirement | Why the requirement matters |
|---|---|
| Read documents in place | Every document stays in iManage, so your firm secures one copy of each document |
| Check access on every query | A lawyer sees only the documents the lawyer can open in iManage, and an ethical wall applies to every answer |
| Search by legal concept | Your lawyers search by client, matter, party, clause and date, not only by the words in a document |
| Answer with a link to the source | Each answer includes a link to the source document in iManage, so a lawyer checks the answer in seconds |
| Work in the AI tools your lawyers use | Claude, ChatGPT and Microsoft 365 Copilot return the same answer from the same documents |
| Log every question | Your firm can show who asked what, what each lawyer saw and which requests the walls refused |
How Reframe searches iManage
- Your firm selects the iManage libraries and workspaces Reframe reads.
- Reframe reads each document through the iManage Work Universal API and files each fact with the client, matter and source page.
- Each lawyer signs in with an iManage account, and Reframe returns only the documents the lawyer can open in iManage.
- Reframe checks the lawyer's ethical walls before any AI tool reads a passage, and refuses a screened request.
- Reframe logs every question, answer and refusal in an append-only audit log.
Why keyword and vector search fall short
Keyword search finds the words in a document, and vector search finds passages that read like the question. Neither search method stores who the parties are, which matter a document belongs to or which amendment governed on a given date.
In Reframe's benchmark on 1,000 fund documents, retrieval search lost 10 percent of its accuracy as the document set grew from 300 to 1,000 documents. With Reframe, the same AI model made 45 percent fewer errors than retrieval search. Read why RAG search loses accuracy as documents grow.
Questions about enterprise search for law firms
Does enterprise search copy documents out of iManage?
Reframe reads documents in place through the iManage Work Universal API, and every document stays in iManage. Reframe keeps a search index and the iManage identifier of each document.
How does Reframe enforce ethical walls in search?
Each lawyer signs in with an iManage account, and Reframe checks the lawyer's access and ethical walls on every query. A security change in iManage applies in Reframe at the next question.
Can lawyers search from Claude, ChatGPT or Copilot?
Yes. Claude, ChatGPT and Microsoft 365 Copilot connect to Reframe over the Model Context Protocol, and each tool signs in as the lawyer who asks.
How does a firm start?
Start with one practice and the iManage workspaces the practice uses. In a 90-minute working session, your lawyers bring the questions they ask, and Reframe moves no documents for the session.