Cloud Migration and AI Readiness

Chapter 3. Top Cloud Migration-Enabled GenAI Scenarios

This chapter focuses on some of the most recurrent AI adoption scenarios. Each section contains the scenario, the key advantages of cloud migration to enable it, and the internal team members who will benefit the most from it. Keep in mind that, as a company, you can be implementing one or several scenarios (e.g., choosing between RAG and fine-tuning, combining both, being a SaaS company but also cloud native).

Let’s now explore those scenarios where cloud migration enables your AI readiness, including the access to advanced technology capabilities in a reduced time and a “pay only for what you use” manner.

Fine-Tuned Models for Specialized Tasks

As we saw in Chapter 1, fine-tuning is one way to “customize” your generative AI developments. In this case, it allows adapting an existing model to perform better on very specialized, specific tasks. This means fine-tuning is less about adding new data to increase the knowledge scope of the models and more about adapting the reasoning process associated with the new tasks. Concretely, fine-tuning can help regular and even small language models (SLMs) to reach similar or even superior levels of performance (an improvement of 25%–50% depending on the models and tasks thanks to fine-tuning) when compared with the latest state-of-the-art GPT models.¹

This means that if you want to adapt an existing baseline model, you need (1) access to fine-tunable versions of those models (keep in ...