Unit 1, Topic 5
In Progress

1.4 Types of Contracts for Adopting AI Software and Services

Unit Progress
0% Complete

Adopting AI software and LLMs involves a variety of contract types and pricing models designed to accommodate different needs and scales of usage. Usage costs for LLMs are typically calculated based on the number of input and output tokens processed, with rates varying by model type and volume of usage. When considering such contracts, it’s crucial to assess your organization’s requirements, estimate usage patterns, and carefully review contract terms, including costs, data handling practices, and support provisions.

Types of contracts

Subscription-Based Licensing

Organizations pay a recurring fee (monthly, annually) to access and use the AI software.

Features:

  • Tiered Plans: Different levels of service (e.g., basic, premium) with varying features and usage limits.
  • User-Based Pricing: Fees may depend on the number of users or seats.
  • Support and Updates: Typically include customer support and regular software updates.
Ideal For: Businesses looking for predictable expenses and ongoing access to the latest features.

Usage-Based (Consumption-Based) Pricing

Charges are based on actual usage metrics, such as the number of API calls, compute hours, or data processed.

Features:

  • Pay-As-You-Go: Only pay for what you use, without upfront commitments.
  • Scalable Costs: Expenses scale with usage, which can be beneficial for variable workloads.
  • Detailed Billing: Usage is tracked and billed accordingly, providing transparency.
Ideal For: Organizations with fluctuating workloads or those starting small and scaling up.

Enterprise Licensing Agreements (ELAs)

Customized contracts negotiated between the software provider and the client.

Features:

  • Fixed Pricing: Agreed-upon pricing for a set period, often with volume discounts.
  • Customized Terms: Tailored features, service levels, and support options.
  • Dedicated Resources: May include dedicated infrastructure or support personnel.
Ideal For: Large organizations with substantial usage requiring personalized terms.

Perpetual Licensing

A one-time purchase grants indefinite use of a specific software version.

Features:

  • Upfront Cost: Higher initial investment with no recurring subscription fees.
  • Maintenance Fees: Optional ongoing fees for support and updates.
  • On-Premises Deployment: Software is often installed on the client’s infrastructure.
Ideal For: Organizations preferring complete control over the software environment.

Open Source Licensing

Software is freely available under an open-source license.

Features:

  • No Licensing Fees: Software can be used, modified, and distributed without cost.
  • Community Support: Reliance on community forums and documentation.
  • Customization: Ability to tailor the software to specific needs.
Ideal For: Entities with technical expertise to manage and customize the software independently.

Additional Considerations in Contracts

Data Usage and Privacy

  • Data Ownership: Clear terms about who owns the data and any outputs generated.
  • Compliance: Adherence to regulations (e.g., GDPR, HIPAA) regarding data handling.
  • Training Data Use: Whether your data may be used to further train or improve the provider’s models.

Service Level Agreements (SLAs)

  • Uptime Guarantees: Commitments on system availability.
  • Performance Metrics: Response times and throughput guarantees.
  • Support Response Times: Expectations for support queries and issue resolution.

Intellectual Property Rights

  • Output Ownership: Rights to the content generated by the AI.
  • Licensing of Generated Content: Any restrictions on how the output can be used or distributed.

Support and Maintenance

  • Technical Support: Levels of support provided (e.g., email, phone, dedicated account manager).
  • Updates and Upgrades: Access to new features and improvements.

Termination Clauses

  • Exit Terms: Conditions under which the contract can be terminated.
  • Data Retrieval: Processes for retrieving your data upon termination.