How accurate is AI contract review compared to a lawyer?

Our AI identifies standard clauses and red flags with 94% accuracy compared to independent review by licensed contract attorneys. For clause identification specifically, accuracy is 98%. However, for high-value contracts (over $50,000), multi-party agreements, or industry-specific regulations, we recommend using AI review as a first pass and then consulting an attorney for the flagged issues. Our users who adopt this hybrid approach report reducing legal costs by 40-60% because the attorney can focus on the specific issues the AI identified rather than reviewing every clause from scratch.

Is my contract data stored or used for AI training?

No. Your contract text is processed in server memory (RAM) only and is never written to disk or stored in any database. When your session ends or after 60 minutes of inactivity, all text is purged from memory. We do not use your contract text to train our AI models. The only data we retain is anonymized, aggregated metadata (e.g., 'non-compete clause, 24 months, creative services' — without any actual contract text) for maintaining industry benchmarks. You can opt out of even this in your account settings.

What types of contracts can the AI review?

The AI handles any English-language contract, but its depth of analysis varies by type. It has the deepest expertise in freelance services agreements, employment contracts, residential leases, and NDAs — these have thousands of benchmark contracts for comparison. It also handles SaaS agreements, commercial leases, vendor contracts, and partnership agreements well. For more specialized contract types like franchise agreements, international contracts, and government contracts, it provides accurate clause identification but less precise benchmarking. The AI supports PDF, DOCX, DOC, RTF, and plain text formats, with documents up to 50 pages.

Can the AI review contracts in languages other than English?

Currently, Contract Review Copilot is optimized for English-language contracts under US law. It has basic familiarity with UK, Canadian, and Australian contract conventions, but its benchmarks and jurisdiction-specific analysis are primarily based on US standards. Contracts in other languages would need to be translated to English first, and the AI would not be able to assess jurisdiction-specific implications for non-US legal systems. We plan to expand language and jurisdiction support in future updates.

How long does a contract review take?

Most contracts are fully analyzed in 2-4 minutes. The breakdown is roughly: 30 seconds for document upload and text extraction, 15-30 seconds for structural parsing, 1-2 minutes for clause identification and categorization, and 30-60 seconds for risk scoring, benchmarking, and generating negotiation language. Very long contracts (40-50 pages) may take up to 5 minutes. Compare this to 3-7 business days for a typical attorney review.

Should I still hire a lawyer if the AI says my contract looks fine?

For standard contracts under $10,000 (typical freelance agreements, simple NDAs, basic service contracts), 83% of our users find the AI analysis sufficient. However, we always recommend consulting an attorney for: contracts involving more than $50,000, employment agreements with equity or complex compensation, commercial leases, business acquisition or partnership agreements, and any contract where a mistake could lead to litigation. Even when the AI gives a green rating across all clauses, there may be strategic considerations (relationship dynamics, negotiation leverage, future implications) that only a human attorney can evaluate.

legal

How AI Contract Review Works

See exactly how our AI analyzes your contracts — no black box

Step-by-Step Process

upload

Upload or Paste Your Contract

You can upload a PDF, Word document, or paste the contract text directly. The AI processes documents up to 50 pages long.

The system extracts text using OCR for scanned documents and preserves formatting for digital contracts. Supported formats include PDF, DOCX, DOC, RTF, and plain text. If the document is scanned or photographed, the OCR engine uses multi-pass recognition to achieve 99.2% character accuracy. The original document structure — headings, numbered clauses, sub-sections — is preserved so the AI can understand hierarchical relationships between clauses. No data is stored after your session ends; all processing happens in memory and is discarded within 60 seconds of session completion.

30 seconds

structure

AI Parses the Document Structure

Before analyzing content, the AI maps the document's architecture — identifying sections, cross-references, definitions, and exhibit attachments.

Many contract risks hide in cross-references. A seemingly reasonable liability cap in Section 7 might be overridden by an indemnification clause in Section 12. The AI builds a dependency graph of all clause relationships, tracks defined terms back to their definitions, and identifies which sections modify or override others. This structural understanding is what separates meaningful contract analysis from simple keyword scanning. The parser also detects boilerplate (standard clauses copied verbatim from templates) versus custom-drafted language, flagging custom clauses for extra scrutiny since they are more likely to contain unusual terms.

15-30 seconds

scan

AI Identifies and Categorizes Key Clauses

The AI scans for 47 standard contract elements including termination clauses, liability limitations, non-compete provisions, indemnification, IP assignment, and payment terms.

Using pattern recognition trained on over 100,000 contracts across 14 industries, the AI categorizes each clause by type and compares it against industry benchmarks. For example, it knows that net-30 payment terms are standard for freelance work but net-60 is common in enterprise SaaS agreements. It also flags missing provisions — if your freelance contract has no kill fee clause, the AI will note that this is a gap that could leave you uncompensated if the client cancels mid-project. Each identified clause is tagged with its category (financial, restrictive, liability, IP, termination, dispute resolution, confidentiality) and its relative importance to the overall agreement.

1-2 minutes

shield

Risk Scoring from Your Perspective

Each clause is evaluated from YOUR side of the contract — whether you are the employee, freelancer, tenant, or business owner — and assigned a risk rating.

This is the most critical step and what differentiates Copilotly from generic AI tools. A broad non-compete is high-risk for an employee but favorable for an employer. An aggressive IP assignment clause hurts a freelancer but protects a client. The AI evaluates every clause from your stated role in the agreement. Risk ratings use a three-tier system: Low (green) means the clause is standard or favorable to you, Medium (yellow) means the clause is slightly unfavorable or contains vague language that could be interpreted against you, and High (red) means the clause significantly disadvantages you or deviates substantially from industry norms. Each rating includes a plain-English explanation of the practical impact — not legal jargon, but real-world consequences like 'this means you cannot freelance for any competitor for 2 years' or 'you would not get paid for work already completed if the project is cancelled.'

30-60 seconds

chart

Benchmark Against Industry Standards

The AI compares your contract's terms against aggregated norms for your specific industry and contract type, highlighting where your terms deviate.

The benchmarking engine draws on anonymized, aggregated data from contract reviews across 14 industries: technology, creative services, healthcare, real estate, retail, manufacturing, professional services, finance, education, hospitality, media, energy, non-profit, and government. For each clause type, the AI knows the typical range. Payment terms for freelance design work cluster around net-15 to net-30; if your contract says net-90, the AI flags the deviation and quantifies its impact (e.g., 'On a $10,000 project, net-90 vs net-30 means you float $10,000 for an additional 60 days, costing you approximately $50-100 in opportunity cost'). The benchmarks are updated quarterly as the system processes more contracts.

20-30 seconds

edit

Generate Negotiation Language

For every medium- and high-risk clause, the AI drafts specific alternative language you can propose to the other party.

The AI does not just say 'this clause is bad' — it gives you the exact words to fix it. For a broad non-compete like 'Designer shall not provide similar services to Client competitors for 24 months,' the AI generates a narrower alternative: 'Designer shall not provide identical services to the following named competitors: [Company A], [Company B] for a period of 6 months following contract termination.' The suggested language is calibrated to be reasonable enough that the other party is likely to accept it. Each suggestion includes a brief explanation of why the proposed change is fair to both sides, giving you talking points for the negotiation. The AI also indicates which clauses are most important to negotiate (pick your battles) versus which are minor enough to accept.

30-45 seconds

report

Deliver Your Complete Review Report

You receive a structured report with an executive summary, clause-by-clause analysis, risk heat map, and prioritized action items.

The final report is organized for quick decision-making. The executive summary gives you a 30-second overview: total clauses analyzed, number of red/yellow/green flags, and a single-sentence recommendation (e.g., 'This contract has 2 high-risk clauses that should be negotiated before signing'). Below that, every analyzed clause is listed with its risk rating, plain-English explanation, industry benchmark comparison, and suggested alternative language. The report ends with a prioritized checklist: which clauses to push back on, which to accept, and which gaps (missing clauses) to request be added. You can ask follow-up questions about any specific clause, request the AI to draft a complete redlined version, or export the analysis for your records.

10 seconds

Sample Output

Clause Risk Assessment

Each identified clause gets a risk rating with a plain-English explanation of what it means for you in practical terms.

Clause 7.2 — Non-Compete Provision

HIGH RISK

This clause restricts you from working in "any related field" for 24 months within 100 miles. This is substantially broader than typical non-competes for freelance agreements (usually 6-12 months, limited to direct competitors).

Practical impact: If you sign this, you could not do freelance design work for any company in a 100-mile radius that could be considered a "related field" — which is vague enough to cover almost any business that uses design services. This could effectively prevent you from freelancing for 2 years.

Enforceability note: Non-competes are unenforceable in California, Colorado, Minnesota, North Dakota, and Oklahoma. In other states, courts often narrow overly broad restrictions, but litigation is expensive.

Suggested revision: "Designer shall not provide identical services to [Named Competitor A] and [Named Competitor B] for a period of 6 months following contract termination."

Missing Clause Alert

The AI does not just analyze what is there — it flags what is missing. Gaps in a contract can be just as risky as bad clauses.

MISSING: Kill Fee / Cancellation Clause

MEDIUM RISK — Gap Detected

This freelance contract has no provision for what happens if the client cancels the project mid-way. Without a kill fee clause, if you complete 80% of a $10,000 project and the client decides to cancel, you have no contractual right to compensation for work already performed.

Industry standard: Most freelance contracts include a kill fee of 25-50% of the remaining contract value, plus full payment for all completed milestones.

Recommended addition: "In the event Client terminates this agreement prior to completion, Client shall pay Designer for all completed milestones at the agreed rates, plus a cancellation fee equal to 25% of the remaining unbilled project value. All work product for paid milestones transfers to Client; unpaid work product remains with Designer."

Executive Summary Overview

The top of every review report gives you a quick-scan summary so you can assess the contract at a glance before diving into details.

Contract Review Summary

Contract type: Freelance Services Agreement

Your role: Service Provider (Freelancer)

Clauses analyzed: 14 of 14

HIGH RISK: 2 clauses — Non-compete (Sec. 7.2), IP Assignment (Sec. 4.1)

MEDIUM RISK: 3 items — Payment terms (Sec. 6.1), Termination asymmetry (Sec. 9), Missing kill fee

STANDARD: 9 clauses — Confidentiality, governing law, dispute resolution, scope of work, and 5 others

Overall assessment: This contract has 2 clauses that significantly disadvantage you and should be negotiated before signing. The IP assignment is overly broad and the non-compete could restrict your ability to freelance. With the suggested revisions, this contract would be within industry norms.

Estimated negotiation priority: 1) IP Assignment 2) Non-Compete 3) Payment Terms

Accuracy & Limitations

94%overall accuracy

Accuracy measured against independent review by 3 licensed contract attorneys across a test set of 500 contracts spanning 12 industries (technology, creative services, real estate, healthcare, retail, professional services, finance, education, manufacturing, hospitality, energy, and non-profit). Each contract was reviewed blind by both the AI and the attorneys, and results were compared for clause identification, risk rating agreement, and quality of suggested revisions. The 94% overall accuracy represents the weighted average across all evaluation criteria.

Strengths

Identifying standard contract clauses and their types (98% accuracy)
Flagging missing critical provisions that should be present for your contract type (96% accuracy)
Detecting cross-reference dependencies between clauses (95% accuracy)
Comparing terms against industry benchmarks (93% accuracy)
Generating enforceable alternative clause language (91% accuracy)

Limitations

Cannot assess enforceability under specific state or local laws with 100% certainty — enforceability depends on jurisdiction, judicial precedent, and case-specific facts
May miss context-specific implications unique to highly specialized industries (e.g., pharmaceutical licensing, defense contracting)
Cannot replace attorney judgment for high-stakes contracts exceeding $50,000 or involving complex multi-party structures
Does not account for verbal agreements or side letters that may modify the written contract
Benchmarks are based on US contract norms; international contracts may have different standards

Sample Conversation

I just got a freelance contract from a new client. Can you review it?

Contract Review Copilot

I would be happy to review your freelance contract. Please paste the text or upload the document. To give you the most useful analysis, I will check for: - Payment terms and timeline - Intellectual property rights and work ownership - Termination and cancellation clauses - Liability and indemnification provisions - Non-compete and exclusivity restrictions - Scope definition and change order process Also, are you the freelancer (service provider) or the client? I will evaluate every clause from your perspective.

I'm the freelancer. Here's the contract: [pastes 3-page freelance services agreement]

Contract Review Copilot

I have analyzed your freelance services agreement. Here is my assessment: **2 High-Risk Clauses Found:** 1. **IP Assignment (Clause 4.1):** This transfers ALL intellectual property to the client, including "preliminary sketches, concepts, and derivative works." This means rejected concepts, drafts they never approved, and potentially your future work that builds on techniques developed during this project all become their property. This is significantly broader than industry standard. 2. **Non-Compete (Clause 7.2):** Restricts you from providing "similar services" to "any competitor" for 24 months within 100 miles. For a freelancer, this is extremely restrictive — it could prevent you from taking on most clients in your area for 2 years. **3 Medium-Risk Items:** 3. **Payment Terms (Clause 6.1):** Net-60 with no late fee. Industry standard for freelance work is net-15 to net-30. On this $8,000 project, you could finish in March and not get paid until May. 4. **Termination Asymmetry (Clause 9):** Client can terminate with 7 days notice. You must give 30 days notice. This asymmetry is common but unfavorable to you. 5. **MISSING — No Kill Fee:** If the client cancels mid-project, there is no provision for compensating you for work already completed. **9 Standard Clauses:** Confidentiality, scope of work, governing law, dispute resolution, and 5 others all look typical for this type of agreement. Would you like me to generate specific alternative language for the high-risk clauses?

Yes, please give me revised language for the IP clause and the non-compete. Also, is the non-compete even enforceable? I'm based in California.

Contract Review Copilot

Good news on the non-compete: **California broadly prohibits non-compete agreements** under Business and Professions Code Section 16600. As of January 2024, AB 1076 further strengthened this by making it illegal for employers to even attempt to enforce non-competes against California-based workers. So this clause is almost certainly unenforceable against you. However, I still recommend addressing it in writing rather than just ignoring it, because: - The contract specifies New York as the governing law jurisdiction (Clause 11.3) - If a dispute arose, the client might argue New York law applies, where non-competes can be enforceable - Having it removed eliminates ambiguity entirely **Suggested revision for IP Assignment (Clause 4.1):** Current: "All work product, including preliminary sketches, concepts, and derivative works, shall be the sole property of Client." Proposed: "Final deliverables accepted by Client in writing shall become the property of Client upon receipt of full payment. Preliminary concepts, rejected drafts, and pre-existing materials created by Designer before or outside this engagement remain the property of Designer. Designer retains the right to display completed work in portfolio materials." **Suggested revision for Non-Compete (Clause 7.2):** Option A (preferred — remove entirely): "Section 7.2 is deleted in its entirety." You can cite California law as the reason. Option B (if client pushes back): "Designer shall not provide services to [Client's Company Name]'s direct competitors, specifically: [Named Company A] and [Named Company B], for a period of 6 months following the completion of services under this agreement." Would you like me to also draft a kill fee clause to add, and suggest better payment terms?

What the AI Actually Checks

When you submit a contract to Contract Review Copilot, the AI evaluates 47 distinct contract elements organized into 7 categories. Understanding exactly what gets checked — and what does not — helps you use the tool effectively and know when to seek additional help.

Financial clauses include payment terms, late fees, expense reimbursement, price adjustment mechanisms, and currency specifications. The AI flags payment timelines that deviate from industry norms and calculates the real-dollar impact. For example, if a contract specifies net-90 payment on a $15,000 project, the AI will note that this is 60 days longer than the industry standard net-30 and that you would be effectively extending $15,000 in credit to the client for 3 months.

Restrictive clauses cover non-compete provisions, non-solicitation agreements, exclusivity requirements, and geographic limitations. These are among the most commonly problematic clauses in freelance and employment contracts. The AI evaluates scope (how broad is the restriction), duration (how long does it last), and geography (how far does it reach), comparing each dimension against typical ranges for your contract type.

Liability and indemnification clauses determine who bears financial responsibility when things go wrong. The AI checks for asymmetric indemnification (where you indemnify the other party but they do not indemnify you), unlimited liability exposure, and missing limitation of liability caps.

Intellectual property clauses govern who owns the work product. The AI checks whether IP assignment is appropriately scoped to deliverables versus all work product, whether you retain portfolio rights, and whether pre-existing IP is carved out from the assignment.

Termination clauses define how either party can end the agreement. The AI flags asymmetric notice periods, termination-for-convenience provisions that favor one side, and missing provisions for compensation upon early termination.

Confidentiality and data clauses are reviewed for scope, duration, and whether the definition of confidential information is reasonable or overbroad. The AI also checks for data handling, retention, and destruction requirements.

Dispute resolution clauses specify governing law, jurisdiction, and whether disputes go to court or arbitration. The AI flags mandatory arbitration clauses (which typically favor the larger party) and governing law provisions that specify jurisdictions far from your location.

How Risk Scoring Works

Every clause identified in your contract receives a risk score based on three independent factors: deviation from industry benchmarks, practical impact severity, and enforceability uncertainty. Understanding this scoring system helps you prioritize which clauses to negotiate and which to accept.

Deviation from benchmarks measures how far a clause strays from the norm for your contract type and industry. The AI maintains benchmark ranges for each clause type across 14 industries. A non-compete of 6 months with named competitors is within the normal range for freelance creative services; a non-compete of 24 months covering "any related field" is a significant deviation. Clauses within the normal range score Low on this factor; moderate deviations score Medium; extreme deviations score High.

Practical impact severity assesses the real-world consequences if the clause is invoked. A slightly unfavorable payment term (net-45 instead of net-30) has low practical impact — you wait an extra 15 days. But an overly broad IP assignment clause could cost you your entire portfolio and restrict your ability to reuse techniques you developed. The AI quantifies impact in dollars, time, and career restriction wherever possible. If a non-compete could prevent you from earning income for 24 months, that is flagged as high severity regardless of whether it deviates from benchmarks.

Enforceability uncertainty captures ambiguity. Vague terms like "related field," "reasonable efforts," or "competitive services" create uncertainty because they could be interpreted broadly or narrowly depending on the jurisdiction and circumstances. The AI flags vague language and explains the range of possible interpretations. Higher ambiguity increases the risk score because you cannot predict how the clause would be applied in practice.

The three factors are weighted to produce a final risk rating. A clause that scores High on all three factors (large deviation, high impact, high ambiguity) gets a red rating. A clause that scores High on one factor but Low on others might receive a yellow rating. The system errs on the side of caution — it is better to flag a clause that turns out to be fine than to miss one that could cause problems.

You can always ask the AI to explain the reasoning behind any specific risk rating. If you disagree or have context the AI does not have (such as a long-standing relationship with the other party), you can factor that into your own decision-making.

Where the AI Gets It Right — and Where It Does Not

We believe transparency about accuracy is more valuable than marketing claims about perfection. Here is an honest breakdown of where Contract Review Copilot excels and where it has known limitations, so you can use it confidently within its strengths and seek additional help for its blind spots.

Strongest performance (95-98% accuracy): Clause identification and categorization is the AI's strongest capability. Given a standard business contract, the AI correctly identifies and categorizes clauses with 98% accuracy. It very rarely misses a clause entirely, and it correctly distinguishes between clause types (e.g., it does not confuse a confidentiality clause with a non-compete, even when they appear in the same section). It also performs well at detecting missing clauses — provisions that should be present for your contract type but are absent.

Good performance (90-95% accuracy): Risk assessment and benchmarking sit in this range. The AI correctly rates clause risk levels about 93% of the time when compared against attorney assessments. Most disagreements are on borderline cases (attorney says Medium, AI says High, or vice versa) rather than gross misclassifications. The AI occasionally over-flags clauses as risky when they are actually standard for a niche industry, because its benchmarks are weighted toward the most common contract types.

Moderate performance (85-90% accuracy): Jurisdiction-specific analysis is the AI's weakest area. Contract law varies significantly by state and country, and while the AI has a general understanding of major jurisdictional differences (e.g., California's non-compete prohibition), it does not have the nuanced understanding of local judicial precedent that a practicing attorney in that jurisdiction would have. For contracts governed by less common jurisdictions or involving cross-border elements, the AI's jurisdictional analysis should be treated as a starting point, not a definitive answer.

When you should still consult a lawyer: For contracts involving more than $50,000, multi-party agreements with complex indemnification chains, intellectual property licensing for patents or trade secrets, mergers and acquisitions, and any contract where a mistake could result in litigation — use the AI review as a first pass and then bring the flagged issues to an attorney. Our users report that this approach typically reduces legal costs by 40-60% because the attorney can focus on the issues the AI identified rather than reviewing every clause from scratch.

What Happens to Your Contract Data

Contracts contain sensitive information — financial terms, business strategies, personal details. Here is exactly what happens to your data when you use Contract Review Copilot, with no vague corporate language.

During analysis: Your contract text is processed in memory on our servers. The AI model reads the text, performs its analysis, and generates the review. This processing typically takes 2-4 minutes. During this time, the contract text exists in server memory (RAM) only — it is not written to disk or stored in any database.

After your session: When your session ends (or after 60 minutes of inactivity), all contract text is purged from memory. We do not retain copies of your contracts. We do not use your contract text to train our AI models. We do not share your contract text with any third party.

What we do retain: We keep anonymized, aggregated metadata for benchmarking purposes. This includes clause types detected (e.g., "non-compete, 24 months, creative services industry") but NOT the actual text of your clauses. This metadata is what allows us to maintain and update our industry benchmarks. You can opt out of even this anonymized data collection in your account settings.

Encryption: All data transmitted between your browser and our servers is encrypted with TLS 1.3. The connection between our servers and the AI model uses encrypted internal networking. At no point does your contract text travel unencrypted.

Compliance: Our data handling practices are compliant with GDPR, CCPA, and SOC 2 Type II standards. If you are a law firm or enterprise user evaluating Copilotly for professional use, we can provide our SOC 2 report and execute a Data Processing Agreement (DPA) or Business Associate Agreement (BAA) as needed. Contact our enterprise team at [email protected] for details.

The bottom line: We process your contract, give you the results, and forget your contract existed. Your sensitive documents are not stored, not used for training, and not accessible to anyone at Copilotly. We make our money from subscriptions, not from data.

AI Contract Review vs. Hiring a Lawyer

This is not an either-or choice. The smartest approach combines both tools for different situations. Here is an honest comparison to help you decide when AI review is sufficient and when you need a human attorney.

Cost comparison: A lawyer charges $200-$500 per hour for contract review, with a typical single-contract review costing $300-$1,000 depending on complexity. Contract Review Copilot is included in the Pro plan at $29/month with unlimited reviews. For a freelancer reviewing 5-10 contracts per year, that is $29/month versus $1,500-$10,000/year in legal fees. However, the cost of NOT having a lawyer review a high-stakes contract can be far higher — a bad employment contract could cost you $50,000+ in restricted earning potential.

Speed comparison: AI review takes 2-4 minutes. A lawyer typically takes 3-7 business days to review a contract and provide a markup. When a client sends you a contract on Friday and wants it signed by Monday, the AI gives you an immediate first read that a lawyer simply cannot match.

Depth comparison: A lawyer brings contextual judgment that AI cannot replicate. They know the opposing counsel's reputation, they understand your business relationship dynamics, they can advise on negotiation strategy (not just clause language), and they can represent you if a dispute arises. The AI provides analysis; a lawyer provides strategy and advocacy.

The recommended hybrid approach: Use AI review for every contract to get an immediate first pass. For standard freelance agreements, NDAs, and simple service contracts under $10,000, the AI review is typically sufficient on its own — 83% of our users report that the AI analysis gave them all the information they needed. For complex contracts, high-value deals, or situations where the AI flags significant risks, take the AI's flagged issues to a lawyer. This focused review costs less (lawyers charge less when they only need to address specific issues rather than review from scratch) and is faster (the lawyer already has a roadmap of concerns). Our users who adopt this hybrid approach report average legal cost savings of 40-60%.

Our honest recommendation: We want you to use our tool appropriately. If you are signing an employment contract with equity, a commercial lease over $50,000, or a business acquisition agreement — please consult an attorney. Use our AI to prepare and identify questions, but get a human professional for the final review. For everything else, Contract Review Copilot gives you a level of analysis that most people have never had access to at an affordable price.

Supported Contract Types and Specialized Knowledge

Contract Review Copilot is not a one-size-fits-all tool. It has specialized knowledge for different contract types, with different benchmark databases and evaluation criteria for each. Here is what it knows best — and where its knowledge is thinner.

Tier 1 — Deep expertise (thousands of benchmark contracts):

Freelance services agreements: The most common contract type our users submit. The AI has deep benchmarks for payment terms, IP assignment, kill fees, revision limits, scope definitions, and non-competes across creative, technical, and consulting freelance work.
Employment contracts: Including offer letters, employment agreements, and separation agreements. The AI evaluates compensation structures, equity terms (vesting schedules, cliff periods, acceleration clauses), non-compete enforceability by state, IP assignment for side projects, and severance provisions.
Residential leases: The AI knows standard terms for all 50 US states, including security deposit limits, rent increase restrictions, maintenance obligations, and early termination penalties. It flags clauses that violate state-specific tenant protection laws.
NDAs and confidentiality agreements: Evaluates scope of confidential information definitions, duration, carve-outs for publicly available information, and hidden non-compete or non-solicitation provisions embedded within NDAs.

Tier 2 — Good expertise (hundreds of benchmark contracts):

SaaS and software licensing agreements
Commercial leases
Vendor and supplier agreements
Partnership and operating agreements
Consulting and advisory agreements

Tier 3 — Basic coverage (general contract analysis without deep benchmarks):

Franchise agreements
Licensing and royalty agreements
International contracts
Government contracts
Construction and building contracts

For Tier 3 contract types, the AI still identifies and categorizes clauses accurately, but its benchmarking and risk assessment may be less precise because it has fewer comparable contracts in its reference database. For these contract types, we especially recommend using the AI review as a starting point and consulting a specialist attorney for the final assessment.

Explore related tools: Legal Copilot for general legal questions, Employment Law Copilot for workplace-specific issues, and Tenant Rights Copilot for rental disputes that go beyond the lease itself.