2 Adoption Models for GenAI

The chapter explains that before governance controls can be designed, an organization must first understand how it is actually adopting generative AI. It frames this as a GenAI posture, noting that many organizations use several postures at once: SaaS tools for convenience, APIs for integration, and self-hosted models for maximum control. The central point is that the adoption model determines where control, dependency, accountability, and risk sit, so governance must be matched to the specific deployment path rather than treated as one-size-fits-all.

It then compares the three main adoption models. SaaS consumers get fast deployment with minimal technical effort, but most controls, data handling, logging, and validation remain in the vendor’s hands, so governance depends heavily on vendor due diligence and contractual protections. API integrators retain more control over the user interface, data flow, logging, and pre- and post-processing, which improves customization and compliance handling, but it also creates shared responsibility and the need for careful coordination with the vendor. Model hosters run the model and surrounding stack themselves, gaining the most control over data residency, tuning, performance, and guardrails, but also taking on full responsibility for security, uptime, supply chain integrity, and regulatory compliance.

The chapter also introduces agentic AI as a cross-cutting development that changes the risk picture in every posture. Unlike ordinary chatbots, agentic systems can plan, observe, use tools, act on external systems, and maintain persistent memory, which makes them more powerful but also more unpredictable and harder to govern. Because agents can take multi-step actions with limited human intervention, the chapter emphasizes risks such as hallucinations, deception, hidden failure propagation, and unsafe autonomy, and concludes with a fictional healthcare case showing how governance needs evolve as an organization moves from SaaS prototyping to API use, self-hosting, and eventually agentic capabilities.

Different types of GenAI adoption. SaaS Consumers have the least control and highest vendor dependency, Model Hosters the highest control and lowest vendor dependency

Decision Tree for Classifying Your GenAI Adoption Model. Use this flowchart to determine your organization’s current GenAI posture and assess relevant security and privacy risks.

SaaS / Application Adoption Model

Shared responsibilities across organizations (client, vendor, LLM provider) in the API Integrator adoption model

Shared responsibilities across organizations (client, vendor, LLM provider) in the Self-Hosted Model adoption model: This architecture assumes the LLM (e.g., Mistral, Llama 3) is deployed on organization-managed infrastructure (e.g., cloud or on-prem). All inference traffic, safety filtering, and post-processing are managed internally. In multi-tenant systems, organizations often share centralized model serving and safety services but must enforce tenant-scoped access controls and storage partitioning for RAG, prompts/secrets, and logs to avoid cross-tenant data exposure.

A high-level architecture of how a multi-agent system works[22]. Agents may use in separate model instances or share a common backbone model.

An illustrative example of the user interface of an AI-powered medical chatbot such as MediAssist, which includes features about diagnosis and appointment booking.

The first Medicare prototype using a SaaS model. It provides advice based on the input provided by the user.

API-Integrator model is now adopted by MediAssist. Behind the scenes, the LLM receives the de-identified patient data to be able to make better recommendations. The UI is hosted at the hospital servers and offers several options such as saving info to the patient chart or sending it to a clinician for review.

Model hoster posture is now adopted by MediAssist. The model is fine-tuned with anonymized patient data and hosted by MediAssist.

FAQ