Gains:
- Ability to recognize distinctive clauses of confidentiality, service, supply, license and distributorship agreements
- Ability to review critical articles specific to the contract type with targeted prompts
- Ability to create type-sensitive checklists for different contract types
“Contract review” is not a single task. Reviewing a confidentiality agreement (NDA) and reviewing a distributorship agreement require very different reflexes; Each type has its own critical items and unique traps. You get the best out of artificial intelligence (AI) when you tell it what type of contract you're working with and what to look for in that type. A general “find risks” instruction produces a species-blind scan; The instruction "this is a one-way NDA, focus on confidentiality period and exceptions" is a targeted and in-depth review. In this unit, we will introduce five common types of commercial contracts and create a type-sensitive prompt and checklist for each.
Let's clarify the terms. NDA (Non-Disclosure Agreement) is the commitment of the parties to keep the information they share confidential; It can be unidirectional (explains one side) or bidirectional (reciprocal). SLA (Service Level Agreement) determines the measurable quality commitments of the service (uptime, response time) and the sanction if not met. License is the permission to use a right (software, trademark, patent); It can be exclusive or non-exclusive. Distributorship is the granting of the right to sell and distribute a product in a certain region. A type-sensitive review is a review that focuses on critical clauses specific to the type of contract.
The Five Types and Their Distinguishing Items
Each type carries a different "risk centre". Giving the AI this map makes the review less superficial.
Contract type
critical items
The most common trap
Privacy (NDA)
Direction (odd/even), duration, exceptions, return/destruction
Indefinite or very broad privacy coverage
Service (SLA)
Service description, level commitments, penalty, termination
Unmeasurable "reasonable effort" statements
Supply/Sales
Price/revision, delivery, defect, delay penalty
Unilateral right to increase prices
Bachelor's degree
Scope, exclusivity, duration, sublicense, IP
Indefinite scope of use, unlimited sublicenses
Distributorship
Territory, exclusivity, target, termination, competition
Stock after non-competition and termination
Non-disclosure agreement (NDA)
The first thing to clarify in an NDA is the direction: is only one party disclosing the information (one-way), or is it mutual (two-way)? This determines the entire balance of obligations.
This is a non-disclosure agreement (NDA). Examine it in this order: 1) Is it one-way or two-way, on which side is the obligation? (with quote)2) How long is the confidentiality period; Is it reasonable or indefinite?3) Are there any exceptions (public information, independent development, court decision)?4) Is the return/destruction of information regulated when the contract ends?5) What are the penal terms and liability? Give each determination with item number and quote; Mark the missing topic as "none".
Service agreement (SLA)
In service contracts, the risk lies in unmeasurable commitments. Phrases like “reasonable efforts will be made” are useless in a dispute.
This is a service/SLA agreement. Focus:- Is the service scope clear and measurable, or ambiguous?- Are level commitments (uptime %, response time) numerical? Are there sanctions (service credit/penalty) if not met? - Termination and exit provisions: data return, transfer support? Mark vague ("reasonable efforts", "to the extent possible") statements in a separate list and suggest measurable alternatives.
Tip: Making the AI say “list vague/unmeasurable statements separately” quickly reveals the most common weakness of service contracts. Numerical commitments such as “within 4 hours” rather than “as quickly as possible” are provable in dispute.
License and distributorship
Scope and exclusivity in licensing, territory, target and post-termination order in distributorship are critical.
It's a [license/distributorship] agreement. Review by type: For licensing: scope (what use), exclusivity, geography, duration, right to sublicense, IP ownership and liability for infringement. For distributorship: territory, exclusivity, sales target and consequence if not met, non-compete, stock/return and customer transfer after termination. Label and quote each item as "balanced / disadvantageous to us / disadvantageous to the other party".
It is not always possible to understand the type of a contract from its title; A text titled "Cooperation Protocol" may actually be a distributorship or license agreement. Therefore, it is useful to have AI diagnose the true nature of the contract before review.
Identify the real type of the contract below by looking at its content, not its title: which commercial contract type (confidentiality, service, supply, license, distributorship or mixed) is closest to it? Base your justification on decisive points (with quotes). If it is a mixed structure, write down which elements it contains separately. Then list which critical items specific to this genre I should prioritize.
Weak Prompt / Strong Prompt
Weak prompt:Find the risks of this contract.
The result: a generic list that clings to every contract, missing the true risk center of the genre. An NDA may focus on the termination penalty and skip the important confidentiality period.
Powerful prompt: [contract type is specified + type-specific critical clause list + pitfall warnings + clause number/citation requirement + balance tags]
Result: A review that focuses on the real risks of the genre, looks for pitfalls, and evaluates the balance. The same pattern goes much deeper with the right guidance.
Creating a Genre-Sensitive Checklist
AI is powerful at producing reusable checklists as well as one-off review. Once you establish a good genre-sensitive checklist, your team will be held to the same standard on every new contract.
Create a reusable review checklist for [supply] contracts. Let there be 12-15 items; Write each item in the form of a yes/no check ("Was the price revision condition regulated by both parties?") Add a "red lines" heading at the end: whatever findings are seen, the signature should be stopped. Assume that a lawyer would approve and institutionalize this.
Three Mini Cases
Case 1 — Indefinite confidentiality. A technology company was about to sign an NDA sent by a supplier. Genre-sensitive NDA review cited that the confidentiality obligation was "indefinite" and that the standard exceptions (public information, independent development) were nonexistent. This was a trap that could make even the company's own independently developed technology count as "infringement" in the future. The period was limited to 5 years and standard exceptions were added.
Case 2 — Unmeasurable SLA. One organization relied on the phrase "high availability provided" in its cloud service agreement. The AI's ambiguous expression screening flagged this; It showed a numerical uptime commitment and no service credit if not met. The clause "99.9% uptime, if it falls below 10% of the monthly fee, credit" was added in the negotiation. In an outage six months later, the institution was able to request a service credit of 240 thousand TL thanks to this article.
Case 3 — Distributorship non-compete. A manufacturer realized through AI's type-sensitive review that in its distributorship contract, there was no right to "not be able to sell competing products for 2 years" and no right to return the stock on hand after termination. These two items together left the distributor with both stock and inactivity in case of termination. During the negotiation, the non-competition ban was reduced to 6 months and a stock repurchase condition was added upon termination.
Common mistakes
- To have it examined without specifying the type. If you don't tell the AI the type of contract, it will miss the true risk center of the type.
- Skipping direction in NDA. Other items cannot be read correctly without determining whether they are one-way or two-way.
- Considering ambiguous expressions normal. Phrases like “reasonable efforts” are weak in conflict; should be made measurable.
- Leaving the scope unclear in the license. Unlimited or indefinite usage/sublicense rights create major IP risk.
- Not institutionalizing the checklist. One-time reviews do not create standards; Make type-sensitive lists persistent.
- The final word on the AI label. The “balanced/disadvantaged” label is a start; People make the decision based on your position.
In summary
Contract review varies by type; Each type has its own critical items and traps. When you tell the AI the type and where to look in that type, a general scan turns into a deep, targeted inspection. Highlight direction and duration in NDA, measurability in service agreement, scope in license, post-termination balance in distributorship; Institutionalize genre-sensitive checklists. AI asks genre-appropriate questions and signals balance; The competent professional decides which balance you will accept.
Application task
Choose two different types of contracts (e.g., an NDA and a service agreement). (1) Run the genre-sensitive review prompt for each and cite critical genre-specific items. (2) List the ambiguous statements in the service contract and have measurable alternatives produced. (3) Create a 12-15 item reusable checklist for one of the species and include a “red lines” section. (4) Verify at least three of the findings from the source material.
checklist
- [ ] Is the type of contract clearly stated on the prompt?
- [ ] Have genre-specific critical items (such as direction/duration in the NDA) been reviewed?
- [ ] Are vague/unquantifiable statements listed separately?
- [ ] Have the scope, exclusivity and post-termination order been considered in the license/distributorship?
- [ ] Is each determination given with item number and citation?
- [ ] Has a genre-sensitive checklist been institutionalized?
- [ ] Balance tags re-evaluated with your own position?