How to Use This Checklist
Use this checklist every time you review a contract that includes AI features — whether it's an AI-native tool, a traditional SaaS product with new AI add-ons, or a renewal where the vendor is introducing AI-related pricing changes.
Work through each section before signing. Check off items as you review them. Anything you can't check off is a conversation to have with the vendor — or a flag for legal.
This checklist covers four areas: Pricing & Cost Predictability, Data Rights & Privacy, Performance & Accountability, and Flexibility & Exit. A Red Flags section at the end highlights the contract language that should stop the process until legal weighs in.
Section 1: Pricing & Cost Predictability
The biggest risk in AI contracts isn't the first-year price. It's what happens at renewal and what happens when usage exceeds your plan.
- Pricing model is clearly defined. Is this per-seat, per-user, credit-based, consumption-based, flat fee, or a hybrid? If it's credit-based, is there a clear conversion rate between credits and actual usage?
- Annual uplift is capped in writing. What's the maximum price increase at renewal? Push for 3–5%. If the vendor won't cap it, request a most-favored-nation (MFN) clause guaranteeing you won't pay more than comparable customers.
- Overage charges are documented and capped. If you exceed your plan (credits, API calls, storage, seats), what happens? Are overages billed automatically, or is there a notification and approval step? Is there a spend cap you can set?
- AI features vs. base product pricing is separated. Is the AI functionality included in your current tier, or is it a separate add-on with its own pricing? If it's included now, can the vendor unbundle it later and charge separately?
- Credit-to-dollar conversion is transparent. For credit-based models: how much does one credit cost? What actions consume credits? Can credit costs change mid-contract? Is there a dashboard to monitor credit usage in real time?
- Pricing model changes require notice. If the vendor transitions from per-seat to consumption pricing (or vice versa), does the contract require advance written notice and your consent? What's the notice period?
- Discount structure is unconditional. If you're receiving a discount, is it tied to purchasing AI add-ons or committing to a longer term? Conditional discounts should be evaluated separately from the base renewal.
- Benchmarking rights are included. Does the contract allow you to benchmark pricing against comparable vendors or industry standards? This is especially important for credit-based models where there are no public price lists to compare against.
Section 2: Data Rights & Privacy
AI tools interact with your data differently than traditional SaaS. The contract needs to reflect that.
- Training data usage is addressed. Does the vendor use your data to train, fine-tune, or improve their AI models? Is this opt-in or opt-out? Is the opt-out available at the contract level, the account level, or the individual user level?
- Opt-out is contractually binding. If you opt out of training data usage, is that commitment written into the contract — or just a settings toggle that the vendor can change in a terms-of-service update?
- Output ownership is clearly assigned. Who owns the content, analysis, or recommendations generated by the AI using your data? Does the vendor retain any rights to AI-generated outputs?
- Data handling on cancellation is defined. What happens to your data if you cancel? Is it deleted within a specific timeframe? Can you export it in a standard format? Is there a transition window?
- Data handling on acquisition is defined. If the vendor is acquired, what happens to your data and your contract terms? Do you have the right to terminate if the acquiring entity has different data practices?
- Data residency and geographic restrictions are documented. Where is your data stored and processed? Does it cross borders? If you have regulatory requirements (GDPR, CCPA, SOC 2, HIPAA), are they addressed explicitly?
- AI content indemnification is included. If the AI generates content that infringes on a third party's intellectual property, who's liable? Does the vendor indemnify you for claims arising from AI-generated outputs?
- Sub-processor transparency is addressed. Does the vendor use third-party AI models or infrastructure (e.g., OpenAI, Anthropic, AWS Bedrock)? If so, are those sub-processors named in the contract, and are they subject to the same data protection terms?
Section 3: Performance & Accountability
AI features are newer, less proven, and more likely to change. The contract should account for that.
- SLA covers AI features specifically. Does the uptime and performance SLA extend to AI-specific functionality, or only to the base platform? If AI features go down, is that covered?
- Service credits are automatic. If the vendor misses SLA targets, do service credits apply automatically — or do you have to request them?
- Audit rights are included. Can you audit or request documentation on how the AI features use your data, what models are being used, and whether your opt-out preferences are being respected?
- Material change notification is required. If the vendor changes the underlying AI model, switches providers, or materially alters the functionality of AI features, are they required to notify you in advance? What's the notice period?
- Feature deprecation protections exist. If the vendor discontinues an AI feature you rely on, what are your remedies? Can you terminate early? Is there a minimum notice period before deprecation?
- Accuracy and reliability standards are addressed. For AI tools where output quality matters (analytics, forecasting, content generation), are there any performance benchmarks or quality commitments in the contract?
Section 4: Flexibility & Exit
AI adoption is unpredictable. Your contract should give you room to adjust.
- Mid-term seat or usage adjustments are allowed. If adoption is lower (or higher) than expected, can you adjust your plan mid-contract? Are there penalties for reducing usage?
- Short-term or pilot options are available. If this is a new AI tool, is there a pilot structure with a shorter commitment and the option to expand? Have you negotiated a walk-away clause if the pilot doesn't meet defined success criteria?
- Cancellation notice period is reasonable. What's the required notice period to cancel or non-renew? Is it 30, 60, or 90 days? Is the notice window clearly documented and not buried in an appendix?
- Auto-renewal terms are acceptable. Does the contract auto-renew? If so, what's the notice window to opt out? Is the auto-renewal at the current rate, or at a new rate the vendor sets?
- Data export and transition provisions exist. If you leave, can you export all your data in a standard format (CSV, JSON, API)? Is there a transition window (30–90 days) after cancellation where you still have access?
- Termination for convenience is available. Can you terminate the contract early without cause? If so, what's the penalty (if any)? This is especially important for first-time AI tool purchases where long-term fit is uncertain.
Red Flags: Escalate to Legal Before Signing
If you encounter any of the following in a contract, pause the process and involve legal counsel. These aren't necessarily dealbreakers, but they require expert review.
- Automatic enrollment in new AI features. The contract allows the vendor to add new AI features to your account — with new pricing — without requiring your approval. Look for language like "features may be added to your plan" or "pricing subject to change upon introduction of new capabilities."
- Unilateral pricing changes tied to "platform updates." The vendor reserves the right to change pricing as part of platform updates, product improvements, or infrastructure changes. This effectively removes your uplift cap.
- Broad IP assignment language. The contract includes language that grants the vendor rights to any content, data, or work product generated within the platform — including AI outputs. Look for phrases like "perpetual, irrevocable license" applied to user-generated or AI-generated content.
- Vague definitions of "AI features." The contract uses broad or undefined language for what constitutes "AI features" or "AI functionality". This creates room for the vendor to reclassify existing features as AI features and apply new pricing.
- No opt-out for data training. The contract doesn't address whether your data is used to train the vendor's models, or the opt-out is buried in a terms-of-service document that can be changed unilaterally.
- Forced arbitration for AI-related disputes. The contract requires mandatory arbitration for any disputes related to AI features, data usage, or output quality — removing your ability to pursue legal action if something goes wrong.
- Model change without notice. The vendor can swap the underlying AI model (e.g., from GPT-4 to a proprietary model) without notifying you. This matters because model changes can affect output quality, cost, and data handling.
Quick Reference: What to Push For
| Contract Area | What "Good" Looks Like | What to Push Back On |
|---|---|---|
| Annual uplift | Capped at 3–5% in writing | "Pricing reviewed annually" with no cap |
| Data training | Opt-out in the contract, not just settings | Default opt-in with no contractual guarantee |
| Overage charges | Notification + approval before billing | Auto-billing with no cap |
| AI output ownership | Customer retains full ownership | Vendor retains license to AI outputs |
| Model changes | 90-day written notice required | No notification obligation |
| Feature deprecation | 6-month notice + early termination right | No protections |
| Credit pricing | Fixed credit cost for contract term | Variable credit cost at vendor discretion |
| Data on cancellation | Export in standard format, 60-day window | "Data deleted upon termination" |
Tips for Using This Checklist
- Don't try to negotiate everything. Prioritize the 3–5 items that matter most to your organization and focus your negotiation energy there. Uplift caps, data training opt-out, and overage protections are usually the highest-impact items.
- Bring this to your kickoff call. Sharing the checklist with the vendor's legal team early in the process signals that you're informed and sets the tone for the negotiation.
- Document what you can't get. If the vendor won't agree to a term, document the risk and get internal sign-off from stakeholders. At least the decision is deliberate.
- Pair this with the AI Pricing Tactics Decoder (Asset 2 in this toolkit) to understand the vendor's pricing strategy before you start reviewing the contract.

