AI has opened new frontiers in various sectors, and the public sector stands to gain significantly from its adoption. So, AI procurement has been top of mind for governments and other public sector organisations.

It makes sense. AI technologies can reportedly revolutionise public services by enhancing efficiency, decision-making accuracy, and citizen engagement. For instance, AI can streamline administrative processes, provide predictive analysis for better policymaking, and offer personalised services to the public. By automating routine tasks, AI frees up valuable human resources, allowing public sector employees to focus on more complex and impactful work. Additionally, AI’s data processing capabilities can lead to more informed decisions, better resource allocation, and improved public services.

With this context in mind, I’ve tailored this step-by-step post for public sector organisations looking to procure AI solutions, ensuring that these benefits are maximised while navigating the complexities of AI integration.

Step 1: Define your AI objectives

  • Identify needs: Clearly articulate what you aim to achieve with AI, whether improving service delivery, automating processes, or enhancing data analytics.
  • Set goals: Establish SMART goals for the AI implementation. “SMART” goals are specific, measurable, achievable, relevant, and time-bound.

Let’s consider an example. A city council might want to use AI for traffic management to reduce congestion. The corresponding SMART goal would be reducing traffic congestion by 30% within a year.

Step 2: Assess organisational readiness

  • Evaluate infrastructure: Check if your organisation’s existing technical infrastructure can support AI.
  • Assess data availability: Ensure the availability and quality of data for AI applications.
  • Skill analysis: Determine if your staff have the necessary skills or require training.

Keeping with AI for traffic management example, you’d evaluate infrastructure by checking if the current IT system can handle real-time traffic data analysis. And to assess the available data, you’d ensure access to traffic flow data from various sources. Further, for skills analysis, you’d assess if your staff needs training on AI traffic management systems.

Step 3: Conduct market research

  • Explore options: Research the AI market to understand the available solutions and technologies.
  • Study cases: Look into case studies of successful AI implementations in similar public sector contexts.

Considering AI traffic management, you’d investigate different AI traffic management systems available in the market, and examine how other cities successfully implemented AI for traffic control.

Step 4: Engage with vendors

  • Initial contact: Reach out to potential AI solution providers.
  • Request demos: Ask for demonstrations to see how these solutions can be applied to your needs.

It would help to reach out to AI solution providers specialising in traffic management and ask them for a demonstration of how their system can adapt to the city’s unique traffic patterns.

Step 5: Address ethical and legal concerns

  • Data privacy compliance: Ensure solutions comply with privacy laws like POPIA or the GDPR.
  • Ethical guidelines: Set guidelines for ethical AI use, considering fairness and transparency.

It’s crucial to ensure the chosen AI system complies with data protection laws regarding citizen personal data. Plus, from an ethics perspective, ensure the AI system’s decision-making process is transparent and non-discriminatory.

Step 6: Draft and issue an RFP

  • Detail requirements: Clearly outline your AI needs and objectives in the Request for Proposals.
  • Encourage innovation: Allow scope for vendors to propose innovative solutions.

In the RFP for a traffic management system, you’d specify the need for a system that can handle real-time traffic data and integrate with existing infrastructure. You’d also invite proposals for advanced, adaptive traffic management AI solutions.

Step 7: Set evaluation criteria

  • Establish metrics: Define clear metrics for evaluating proposals, including technical capability, cost, and vendor experience.
  • Consider post-support: Factor in the level of support and training the vendor offers post-implementation.

For the AI traffic management system, criteria might include system accuracy, cost, compatibility with existing infrastructure, and vendor experience in similar projects. When considering support, look for vendors offering comprehensive training and technical support, so you don’t have to go to different vendors to support the same AI system.

Step 8: Pilot testing

  • Implement a pilot: Start with a pilot project to test the AI solution in a controlled environment.
  • Evaluate and adapt: Use the pilot to assess effectiveness and make necessary adjustments.

For instance, a pilot project would test the AI system in a small area of the city first, then assess the system’s impact on traffic flow and make adjustments before city-wide implementation.

Step 9: Plan for continuous learning

  • Expect evolution: Understand that AI solutions will evolve and require updates.
  • Invest in training: Ensure ongoing staff training to keep up with AI advancements.

It would be wise to prepare for regular system updates based on the latest AI advancements. So I’d recommend providing ongoing training for staff on system updates and new features.

Need help with AI procurement?

  • Understand the nuts and bolts of AI procurement by asking us to train you or your procurement team.
  • Procure the right AI vendors by asking us to draft or review your RFP documents.
  • Know how the world’s public sectors are procuring AI by asking us to explain the best practice guidelines to you.