Planning and procurement strategy tips

Two types of planning are required for successfully adopting and purchasing an AI solution: project planning, to figure out the broad technology strategy, which is usually led by the agency/department project team or sometimes IT, and procurement planning, to determine which products/services/licenses have to be purchased as part of this project, which is led by procurement.

This section mainly focuses on procurement planning. However, at the end of this section, we also include important considerations for other involved teams during the planning of the broader project.

One valuable tool for procurement planning is a procurement strategy. A procurement strategy is a public-facing plan that explains how your state or local government will buy and deliver a specific technology to achieve measurable outcomes for residents and/or staff.

A procurement strategy should articulate the problem you are trying to address in the context of your organization’s wider strategic objectives and priorities, and the outcomes you are seeking to achieve in plain language. This analysis should include a clear assessment of market solutions versus custom development, explain your chosen delivery approach, your sourcing model (including solicitation and contracts types, terms, timelines) and commercial terms up front. This helps set shared expectations for public agencies, vendors, oversight bodies, and the public, and is particularly important when procuring mission-critical technology.

Early on you’ll need to decide whether to pursue off-the-shelf AI, or customize or build your AI solution. You may also consider reusing existing open-source components as a delivery approach, or in combination with other methods. Your strategy should clearly justify this choice.

Who and how to engage

Whom you gather input from as you develop and refine your strategy is as important as the content itself. Key stakeholders include:

• Your internal experts: You should speak to the agency or department buying the technology, and your technical, finance, and legal experts. Gathering this feedback early will help generate alignment around what you are buying, how you will buy and pay for it, and what to consider during contract implementation with your vendors.
• Your organization’s employees: Seek to empower public servants’ voices throughout the procurement process by engaging with public sector commissioners and trade union representatives.
• Suppliers: Strong, proactive supplier engagement is especially important for AI. AI technology and vendor capacity are changing very fast, so understanding what the market is capable of, now and in the near future, is important for many aspects of your procurement strategy. Potential engagement methods range from attending conferences or vendor fairs, to issuing RFIs and conducting one-to-many and one-to-one supplier sessions, to innovative, challenge-based procurements.
• The public: By engaging the public during the design of your procurement strategy, you can promote organizational transparency and accountability, and build trust with your residents. Public participation is especially important for any AI technology that the public might interact with directly as an end user, or used to influence public sector service delivery.

In addition to your key stakeholders, you may also consider gathering input from organizations that work on AI- and procurement-related topics for the public sector, such as professional associations, nonprofits, or academia.

How you engage your stakeholders matters, too. By taking a collaborative and participatory approach, you can develop a strong shared understanding of the problem you are trying to solve, your objectives, and outcomes.

Thinking through your budget

Identify your funding sources, and which team within your organization will own the product budget. Depending upon the scope of your purchase, you may have multiple teams responsible for different budget lines.

It’s important to recognize that AI solutions typically use different pricing models than traditional software, which have implications for your costs. Think carefully with your IT/data teams around what realistic costs look like, both upfront and for appropriate maintenance.

Common AI pricing models include:

• Usage-based pricing: Many AI services charge per API call, document processed, or query made
• Subscription models: Monthly/annual fees often based on number of users or processing volume
• Hybrid approaches: Combination of base fees plus usage charges
• Custom enterprise pricing: For large implementations or specialized requirements

Key considerations for procurement around pricing include:

• Costs can scale rapidly with usage, making it especially important to understand volume projections.
• Training or customizing models may involve significant upfront costs.
• Data preparation, integration, and ongoing human oversight all have cost implications.
• Consider pilot pricing vs. full-scale implementation costs when budgeting.