Define AI technical requirements

- The second phase of our framework Survey involves conducting a feasibility assessment. It's about looking closely at your business and understanding your organization's current state of AI readiness. This starts with the technical requirements of any AI solution that you plan to adopt. Whether you're using SaaS-based AI products, fine tuning models on your own infrastructure, or accessing proprietary ones like Gemini, Claude, or GBT via an API, technical requirements are an essential consideration. In this video, I'll walk you through how to draft an AI technical requirements document that outlines the technology and infrastructure needed to support your generative AI initiatives. Your AI technical requirements document has twin objectives. First, it audits current organizational capabilities related to data, infrastructure, software, et cetera. And secondly, it defines what is needed to build or adopt preferred solutions. In crafting your AI technical requirements document, here are some elements to review. Firstly, audit your compute infrastructure. If you are building a custom solution, you must decide between running your own models or using APIs. Running your models on your own infrastructure may provide greater control, but demands comprehensive infrastructure management, including sophisticated hardware and skilled personnel. Alternatively, using a managed service or API will provide access to more advanced models and allow you to focus on your product development. Your choice should align with your organization's needs and capabilities, balancing both domain expertise with cost efficiency. Next, recognize that a modern data foundation is an absolute prerequisite for effectively leveraging generative AI. You'll need to catalog your data sources, quality, and security, and assess your organization's data availability, governance policies, and preparedness. You should also consider how seamlessly your AI solution integrates with existing systems and their scalability. As a foundational technology, you want your AI to evolve alongside your business. Next, environmental impacts should not be overlooked. The energy requirements of AI models, especially when developing your own, can significantly affect your organization's carbon footprint. Be sure to assess this. Lastly, it's crucial to incorporate a comprehensive budget estimate in your AI technical requirements document. This should include all potential costs such as licensing fees for proprietary AI models or software, and the expenses associated with infrastructure development or upgrades. Additionally, consider the ongoing operational costs such as technical support, data management, and any necessary staff training or hiring. In summary, defining the AI technical requirements is essential for successful implementation and adoption, whether you're running models on your own infrastructure or leveraging API-based services. So I challenge you take some time after this video and thoroughly document and assess your needs across data, infrastructure, software, processes, and costs to ensure your AI initiatives are technically sound and strategically aligned. We'll build on this document in the next video.