In the PMI-CP in Managing AI guidance, early AI project work includes confirming that the data foundation is viable before committing to specific tools or architectures. For AI initiatives, data is the primary constraint: if the right data does not exist, is incomplete, or is of low quality, no choice of technology will rescue the solution. Therefore, before assessing tooling gaps or even detailing the technology stack, teams are expected to verify the availability, accessibility, and quality of the required data for the intended use case.

PMI-CPMAI describes data readiness activities such as identifying key data sources, profiling them for completeness and consistency, assessing coverage of relevant populations and time periods, and checking for legal and regulatory constraints around access and use. Only after this verification can the team meaningfully evaluate whether existing platforms, infrastructure, and tools are sufficient, and then identify gaps.

Assessing team expertise or procuring tools are important, but they follow from the prior understanding of what data exists and what is needed for the model. Thus, the first step the project team should complete to ensure they have what they need for AI development is to verify the availability and quality of the required data.