select those activities that are critical to your AI strategy and use cases — and then sequence them properly, from initial to advanced.

Approach goal setting with a succinct statement of the choices that your organization will make in order to win. This starts by defining your AI ambitions — that is the strategic impact that you hope to create with AI, in alignment with the business strategy. The initial AI strategy informs the adoption goals for the roadmap and the priorities for the use-case portfolio. Among more advanced activities are establishing a process to refine the strategy, including the measurement of its success.

AI value is directly realized via a portfolio of AI initiatives. You will typically start by prioritizing a set of initial use cases, running pilots, and tracking and demonstrating their business value. More advanced activities include creating a portfolio of AI products, where the focus is not on delivering one-off projects, but on ongoing value creation that continuously evolves with customer needs and changing technology.

Learn how a Gartner Executive Partner can help. As a client, you can work directly with an EP mentor to build your custom AI roadmap.

As AI scales, your organization will need to evolve too. At the start, you need a resourcing plan that responds to the needs of your initial AI use cases and strategy, including whether you will seek to fill key capability gaps internally or externally. On the internal side, organizations typically start by establishing a community of practice that brings together stakeholders interested in AI and/or dedicated AI teams to focus on a limited set of high-priority activities. This can then evolve into a target operating model designed to further scale AI in the organization. On the external side, organizations often initially form a limited number of external partnerships and later formalize a process to manage these relationships.

AI represents a big change for the workforce: Employees will need to be upskilled, some jobs will need to be redesigned and the culture will need to adapt. The first typical step is to create a workforce plan that identifies the talent implications of AI, the current talent gaps and how they will be addressed. This generally evolves into a process to continuously review roles, a change management plan and a protocol for evaluating the impact of AI in the workforce.

AI comes with many risks that need to be governed from the outset. The typical journey starts with identifying the key AI risks and establishing the initial principles, policies and enforcement processes to manage and mitigate those risks, including ethical AI issues. AI leaders can then look to formalize an AI governance structure that defines decision rights and establish a broader AI governance operating model. More advanced activities include piloting governance tooling and setting up AI literacy programs to educate employees more broadly on AI governance.

You need a solid technical foundation to ensure your organization’s AI is both reliable and scalable. Initially, organizations need to define build versus buy criteria for their use cases, set up a sandbox environment to experiment, and start identifying design patterns and reference architecture to promote reusability. They also need to carefully select vendors to support initial use cases, then evolve this to a more cohesive AI vendor strategy. More advanced activities focus on establishing a ModelOps practice, including AI observability, UI/UX and FinOps best practices, as well as standing up AI platform engineering.

Data is a vital component of the vast majority of AI use cases. However, AI-ready data requires a different set of capabilities than traditional data management. AI leaders typically start by understanding their data readiness for the initial set of AI use cases, identifying key requirements and implementing a plan to prepare their data. Then, they seek broader buy-in for the longer-term investments required to evolve their data capabilities for AI and adapt their data governance, including data quality and metadata practices. AI also requires capabilities to visualize and analyze data, as well as data observability to monitor data in production.