Document Review
Discovery Agent can automatically identify document types and configure itself to extract data from contracts and third-party paper using artificial intelligence and machine learning to compare scanned text to provision models. Reviewers monitor the results, verifying and correcting them where necessary. Different provision models extract different types of data and present different workflows.
Reviewing Imported Clauses and Fields
You can review and correct the clauses and fields extracted from imported documents. Discovery Agent converts all scanned or uploaded documents to PDF format for scanning and review. Provisions and other significant text are highlighted for review in an in-app PDF reader, and each is displayed in the column to the right in the order it appears in the document. For each detected clause or field you review, you can accept the extract, reject the extract, edit the extract, or flag the extract for review.
Reviewing Imported Tables
You can review and correct tables extracted from imported documents.This review process is similar to that for clause and field extractions, though the table form requires extra considerations. For each detected table you review, you can accept, reject, edit, or flag the extract for review by line or by table (bulk review).
Searching Document Text
While reviewing a document under extraction, you can search the document text.
Document Summary
A document summary is located under the Review tab.
The AI engine generates a summary field for all documents, except those ingested as OCR-only. A document under summary is indicated as a blank in the review interface with a spinner running. You can click the pencil icon to edit the generated summary.
- The document is summarized when you open it for review.
- Longer documents take longer to summarize.
If there is a summary error, you are warned and asked whether to complete the review when you attempt to save. If you affirm, the review is saved with no summary. If you decline, the save aborts. The Summary tab offers a Regenerate button to retry the summary or you can enter one by hand.
Use Case
A general counsel with a pre-negotiation briefing in 15 minutes needs to understand the key terms of a 120-page distributor agreement without reading it. Opening the document in Discovery Agent, they read the two-paragraph AI-generated summary under the Summary tab and then skim the risk flags in Insights. The counsel walks into the meeting informed, having spent four minutes in Discovery Agent, rather than four hours with the original contract.
Searching Provisions
Adding Unmapped Fields and Clauses
You can map new fields and clauses defined in CLM that are suggested by Discovery Agent from the review page.
If the administrator has enabled it, you can extend the provision models mapped for a document you are reviewing as suggested by Discovery Agent. When you import a document, Discovery Agent finds CLM fields and clauses that have been mapped and indicates which ones it has and has not found. On your request, it can display provisions that have not yet been mapped as suggestions. This feature works for fields and clauses, but not for tables or obligations.
You can review and add unmapped fields and clauses from the extraction to the worksheet before saving them to CLM. Adding an unmapped value maps the subject passage to the provision map, tying it to the fields, clauses, or values available in the CLM schema. Subsequent imports will include the new provision models and use them in extractions; however, this mapping is not retroactively applied to already-imported documents.
Use Case
A paralegal reviewing a newly onboarded software license agreement notices that it contains a “Source Code Escrow” clause, a provision that is absent from the standard software agreement worksheet. Rather than flagging it for later, she clicks Add Unassigned Clauses to map it to an existing CLM clause field in real time. The next import of a software license using this worksheet will include looking for this clause provision model as part of its review.
Adding New Clauses from Reviewed Documents
You can create new clause provision models based on passages you highlight while reviewing.
- Your selected passage is now a clause provision model.
Your new clause is entered into CLM.
Details Tab
The Details tab, adjacent to the Review tab in the document review interface, contains these fields:
- Document Name
- Record Type
- Worksheet Name
- Review Level
- Review completed by
- Owner
- Assignee
- Document comment
The Document Name, Worksheet Name, Review Level, “Review completed by”, and Owner fields are populated from the back end. You can select the record type and assignee and enter comments in the “Document comment” box.
Editing an Extracted Provision
Deleting One or More Provisions
You can delete an unsuitable or inaccurately extracted provision from your results.
- From the Project Dashboard or the Home screen, open a Ready for Review document as described in Reviewing Imported Clauses and Fields.
- If you have opened an erroneous extraction that you will not correct as described in Editing an Extracted Provision, and would prefer to delete altogether, click the More (
) icon to the right of the provision tile in the Review column at right and click the resulting pop-up trash-can (
) icon.
- Click DELETE in the confirmation pop-up.
- Click the checkboxes to the left of the provisions you will delete.
- Click the trash-can (
) icon above the provisions in the Provisions Found header.
- Click YES, DELETE in the confirmation pop-up.
AiMe Assistant
AiMe Assistant is an AI-powered virtual assistant designed to boost productivity and efficiency when working with a contract document through its entire life cycle. It uses machine learning (ML) and natural language processing (NLP) to analyze and generate content. This tool enables users working on contract documents, supporting documents, and global and other documents to summarize them and to get answers about them.
AiMe:
- Answers questions about such documents as contracts, their supporting documents, and global documents.
- Keeps a history of questions users asked it in a document.
- Summarizes contracts, supporting documents, and global and other documents.
Use Case
Used properly, AiMe Assistant can make the most of your time when analyzing contract documents in your workflow.
With 30 minutes left before a renegotiation call with a key supplier, a procurement manager opens the 87-page supply agreement in Discovery Agent and asks AiMe: "What are our volume commitments and what are the penalties for missing them?" and "Does this contract auto-renew?" AiMe returns precise answers to these natural-language questions, with citations in under a minute, saving considerable time and stress to search the provisions for the right fields.
Accessing AiMe Assistant
Discovery Agent can provide access to AiMe Assistant, a conversational AI that can answer questions about documents.
Asking AiMe a Question
You have opened AiMe from CLM, X-Author for Contracts Advanced, Contracts for Salesforce, or Discovery Agent.
Use Case: Verifying a Clause or Field
- The user has uploaded or selected a contract document in a supported (PDF, DOCX) format.
- AiMe has processed and indexed the document .
- The user has appropriate permissions to view the document content.
- The AiMe is available and responsive.
The user, a lawyer, contract manager, procurement specialist, or sales operations manager, wants to verify quickly whether a specific clause (e.g., indemnification, limitation of liability, termination for convenience) or field (e.g., effective date, governing law, notice period) exists in a contract document and if so, retrieve its exact wording or value.
- The user opens the contract document from the parent interface (CLM, Discovery Agent, etc.)
- The user enters a query in the chat box, such as: “Does this contract contain an indemnification clause?”, “What is the governing law in this agreement?”, or “Is there a limitation of liability provision?”
- AiMe processes the query and searches the document for relevant content, returning a response indicating whether the clause or field exists, along with the summarized text.
- The user reviews the extracted clause content and may follow it to the source location.
- A clear yes/no confirmation of the clause’s or field’s presence.
- A text summary of the identified clause or field value.
Alternatives or Exceptions
Clause not found: AiMe responds with “No [clause type] was found in this document”.
- Use specific clause names (“limitation of liability” rather than “liability issues”) for more accurate results.
- For fields, specify the exact field name as it might appear in the document (“Effective Date” rather than “term commencement”, for example).
- Review the full clause in context before making decisions, as the excerpt may not capture all relevant conditions.
Use Case: Asking “Yes or No” Contract Questions
- The user has uploaded or selected a contract document in a supported format.
- The document has been processed.
- The question can be answered based on information contained in the document.
- The user understands that the response is based on AI interpretation and may require verification.
A lawyer, contract reviewer, compliance officer, or business stakeholder wants a yes or no answer to a specific question about contract terms, conditions, or provisions to support rapid decision-making during contract review or negotiation.
- The user opens the contract document from the parent interface (CLM, Discovery Agent, etc.)
- The user enters a yes/no question in natural language, such as: “Does this agreement allow automatic renewal?”, “Is the vendor liable for consequential damages?”, or “Can either party terminate for convenience?”.
- AiMe analyzes the document and returns a clear “Yes” or “No” response, followed by the supporting text from the contract that justifies the answer.
- The user reviews the supporting evidence, proceeding with their own review or decision.
Use Case: Summarizing a Document or Clauses
- The user has uploaded or selected a contract document in a supported format.
- The document has been fully processed and indexed.
- The user has identified the specific focus area(s) or clauses they want summarized.
- For multi-clause summary, the user can specify which clauses to include.
A lawyer, contract manager, executive reviewer, or deal desk analyst wants to accelerate their contract interpretation and review with a tailored summary of the entire contract focusing on specific user-defined aspects, or a consolidated summary of several selected clauses.
- The user opens the contract document in the AiMe interface.
- The user requests a summary with custom guidance, such as: “Summarize this contract, focusing on financial obligations and payment terms,” “Provide a summary of the termination, renewal, and assignment clauses,” or “Give me an executive summary highlighting key risks and unusual provisions.”
- AiMe processes the document with the specified focus on the requested clauses, generating a structured summary that addresses the user’s specific areas of interest, organized by topic or clause.
- The user reviews the summary, which includes references to source locations for each summarized point.
Use Case: Extracting Obligations
- The user has uploaded or selected a contract document in a supported format.
- The document contains identifiable party names or roles.
- The contract includes obligation-related language (shall, must, will, agrees to, commits to, etc.)
A contract manager, compliance officer, project manager, or legal operations specialist wants to identify and extract all obligations (duties, responsibilities, commitments, deliverables, and deadlines) for each party in the contract to support obligation tracking, compliance monitoring, and project planning. Such a user might want to filter obligations by party, type, or time frame.
- The user opens the contract document from the parent interface (CLM, Discovery Agent, etc.)
- The user enters an obligation extraction request, such as: “Extract all obligations from this contract”, “What are the vendor’s obligations under this agreement?”, “List all deliverables and their due dates”, or “What must our company do under this contract?”
- AiMe scans the document for obligation language and returns a structured list of obligations.
- The user reviews the extracted obligations and can export or transfer them to an obligation management system.
Use Case: Doing Math with Table Data
- The user has uploaded or selected a contract document containing one or more tables with numeric data.
- The table structure is simple and clearly formatted (standard rows and columns).
- Numeric values in the table are recognizable (not embedded in complex text or images).
- AiMe has parsed the table content during document processing.
A financial analyst, procurement specialist, contract manager, or commercial operations manager has to perform basic calculations (sum, difference, average, count, minimum/maximum, and ratio) on numerical data from tables in the contract document to analyze pricing, quantities, or other numeric terms.
- The user opens the contract document from the parent interface (CLM, Discovery Agent, etc.)
- The user identifies a table in the document (e.g., pricing table, fee schedule, quantity list).
- The user enters a calculation request referencing the table, such as: “What is the total of all line items in the pricing table?”, “Calculate the average unit price from the fee schedule”, “What is the maximum quantity listed in the order table?”, or “Add the values in the Annual Fee column.”
- AiMe identifies the relevant table, parses the numeric data, performs the requested calculation, and returns the calculated result.
Best Practices
Rules to follow for better results
Do
- Ask direct, factual questions based on the content present in the contract.
Example: “What is the governing law for this agreement?”
- Use AiMe to summarize large sections or entire agreements.
- Phrase your prompts in a clear, natural tone. Specific, business-focused inquiries yield the best results.
- Validate AiMe’s outputs before using them in internal or external communications.
- Use AiMe on contract documents that are either finalized or under-negotiation, provided they are uploaded to the Agreement Documents list in a readable format.
- Ask your AiMe admin to create shortcut prompts for frequently asked queries.
- Use AiMe to analyze the current document only. It cannot reference or connect to external files.
- Use AiMe to help non-legal stakeholders (sales, procurement, finance) navigate dense legal language.
- Remember that AiMe is a summary and insight assistant, not a legal decision-maker.
- Report recurring issues or gaps in responses to Conga.
Don’t
- Request opinions, recommendations, or predictive analysis.
Example: “Will this clause protect us in future disputes?”
- Use vague or broad prompts.
Example: “What’s important here?”
- Ask several questions in a single prompt: AiMe may hallucinate (Ask questions separately).
Example: “Give me the effective date, term length, governing law, payment terms, liability cap and notice period.”
- Ask AiMe to perform complex mathematical operations or multiple operations in a single prompt. AiMe can perform basic operations on numbers in a simple table, such as sum, difference, average, count, min/max, or ratio.
- Use AiMe on unstructured or incomplete drafts that lack meaningful contract content.
- Expect AiMe to red-line, edit, or compare clauses. Use Redline Agent or Search Agent for those tasks.
- Ask AiMe to analyze multiple documents at once. AiMe works on one document at a time.
- Ask AiMe to search extracted metadata. AiMe can only process the uploaded document.
- Assume the output to be exhaustive or 100% accurate. You must review critical terms manually.
- Expect AiMe to support non-English contract review unless Conga has confirmed the language to be supported.
- Ask Copilot to review files that are larger than 12 MB or 200 pages.
Giving Feedback
Your feedback helps AiMe work better.
When AiMe returns an answer, you can assess the quality of its response using the thumbs-up/thumbs-down feedback icons. Clicking thumbs-up returns positive feedback to the AI engine. The assumptions it made to understand your prompt and generate the response are given more weight in the language model.
AiMe also interprets copying the results to the clipboard using the copy button () as a thumbs-up response.
AiMe’s AI engine uses your feedback to inform future responses.
Reviewing the AI Chat History
Recall recently asked questions.
You have opened AiMe from CLM, X-Author for Contracts Advanced, Contracts for Salesforce, or Discovery Agent.
Generating a Document Summary
You have opened AiMe from CLM, X-Author for Contracts Advanced, Contracts for Salesforce, or Discovery Agent.
- The Contract Summary section highlights the key points related to the document.
- The Brief Summary section offers a concise overview of the document.
Reviewing a Document’s History
You can review the history of any fields ingested into the system (in a Ready for Review, Ready for Approval, or Complete status). The Review History contains details about which fields were changed, who changed them, how they changed them (by creating, updating, or deleting), when they changed the fields, and the contents of the field.
During an external audit, a compliance officer must demonstrate that a vendor agreement’s payment terms were extracted accurately and approved by two separate reviewers. Opening Review History and exporting the time-stamped change log shows who extracted the value, who edited it, and who approved it, in a format acceptable to the auditor.
- From the home screen, click the name of an uploaded document.
- Click the Review History tab.
