Worried You've Missed a Clause? Here's How AI Detects Risks in Contracts

What Clauses Can Be Detected Automatically?

AI Automationcan help Legal find potentially risky clauses by reading the contract content, comparing it to the company template, and flagging sections that need to be checked.

• Liability and indemnitywhich differs from company standards.

• Termination clausewith too short a notice period.

• Payment termswhich is not in accordance with the agreement.

• Automatic renewalwhich was not immediately apparent during the initial review.

• Confidentiality dan data protectionwhich is incomplete or changed.

For example, the company has a template with payment terms30 days, but the vendor contract includes60 days. With AI Automation, these differences can be marked before the contract enters the approval stage.

How Does AI Determine Which Clauses Need to be Checked?

Detecting risky clauses should not rely solely on keyword searches. What is more important is comparing the contents of the document withinternal rules and standards.

Workflows that can be used:

Contract Upload → Text Extraction → Clause Detection → Rule Check → Risk Flag → Legal Review

With AI Automation, each stage can be made part of a scalable workflow.

1. Identify Important Clauses

The system can search for clauses such as:

• Limitation of Liability

• Indemnification

• Termination

• Confidentiality

• Payment Terms

• Governing Law

• Data Protection

• Renewal

However, just finding a clause does not mean that the clause is problematic.

For example, the word“liability”found in 5 sections. The system still needs to look at the context and value of the stated obligations.

2. Compare with Template

With AI Automation, the vendor draft can be compared with a standard contract template.

Example:

Standard:Maximum liability is 100% of the contract value.

Draft:Maximum liability is 20% of the contract value.

The change is worth flagging because the vendor liability threshold is much lower.

3. Use Risk Scoring

Risk scoring helps Legal determine which ones to check first.

Example:

Low Risk: 0–30

Medium Risk: 31–60

High Risk: 61–100

Weights can be made based on company needs.

For example:

• Liability: 30%

• Indemnity: 20%

• Termination: 15%

• Data Protection: 20%

• Payment Terms: 15%

With AI Automation, the scoring results can be used to determine the review path.

4. Relate Risk to Contract Value

The same clause can have different impacts depending on the value of the contract.

Contract worthRp. 20 millionand contracts worthRp. 2 billiondoes not always require the same treatment.

For example, the company determines:

Risk High + Contract Value >Rp1 Miliar → Senior Legal Review

With AI Automation, this rule can be a trigger for escalation.

5. Show the Reason Behind the Flag

Don't just label things“High Risk”.

Legal needs to know the cause.

Example results:

Risk: High

Reason: Liability cap differs from template

Source: Section 8, Limitation of Liability

With AI Automation, the results can help the reviewer go directly to the relevant section of the contract.

This is much more useful than asking Legal to reread 30 pages just to find one clause change.

Practical Steps Before Implementing It

• Determine5–10 clausesthe most important.

• Prepare a standard contract template.

• Determine the conditions that are considered risky.

• Create categories Low, Medium, and High.

• Determine the risk scoring weight.

• Set the contract value threshold.

• Make sure each flag has a source clause.

• Implement role-based access.

• Keep an audit trail.

• Test the workflow with20–50 contracts.

• Review the results with Legal before widespread use.

• Use legal software for companies and officially licensed software.

It's important to remember that automatic detection doesn't mean AI can declare a clause to be illegal or a contract to be safe. Its primary function isfind patterns and differences worth examining, so that Legal can allocate time to the most important parts.

FAQ

1. What clauses are best suited for AI to detect?

Liability, indemnity, termination, payment terms, confidentiality, renewal, governing law, and data protection are commonly used examples.

2. Can AI recognize different clauses from a template?

Can. AI Automationcan help compare the contents of the draft with predetermined standards.

3. Can AI immediately declare the clause dangerous?

It's best not to. The system flags based on rules and patterns. Legal authorities still need to assess the legal implications.

4. Is it mandatory to use numbers for risk scoring?

No. Categories like Low, Medium, and High can also be used. Numbers simply help make priorities more measurable.

5. Can the contract value be used in risk assessment?

Yes. The contract value can be an additional parameter for determining the level of review and approval.

6. What if the contract formats are different?

Documents can first be processed through text extraction or OCR. The system can then search for and compare relevant information.

7. Do detection results have to include the source page?

Highly recommended. Reviewers need to be able to double-check the original clause before making a decision.

8. Can this process be connected to approval?

Can. AI Automationcan direct documents based on risk score, clause type, contract value, or company approval rules.