Revolutionizing Public Sector Service Delivery With Microsoft Azure OpenAI Service

AI in its different forms, both predictive and generative, improves service delivery internally across government institutions as well as externally with constituents. Internally, AI enhances decision-making and enables governments to anticipate needs and allocate resources more effectively. Externally, AI encourages citizen engagement by making it more interactive and personalized. AI insights guide policy development and ensure leaders create and assess more data-driven options.

Forrester research found that 35% of government and public sector decision-makers reported that their organization is planning to implement AI infrastructure in the next 12 months.¹ These decision-makers also reported the AI use cases that were considered important for their organization in the next 12 months, including:

• Improving data management (36%).
• Delivering AI-enhanced constituent experiences (32%).
• Improved operational efficiency and effectiveness (33%).
• Increased automation of internal processes (33%).

Microsoft commissioned Forrester Consulting to interview 20 representatives at 16 organizations and conduct a Total Economic Impact™ (TEI) study to better understand the benefits, costs, and risks associated with the investment in Microsoft Azure OpenAI Service. 2 In the full TEI study, Forrester found that an industry-agnostic composite organization would see risk-adjusted financial benefits ranging between $45.9 million and $197.4 million over three years.

*For the composite organization presented in the full TEI study

This abstract will focus on the government sector’s use of Microsoft Azure OpenAI Service and its value to their organizations.

As part of the broader TEI study, Forrester interviewed representatives with direct experience implementing Azure OpenAI Service at government agencies:

• An AI lead supporting government agencies in APAC that serviced between 90 and 130 million constituents across eight different government departments.
• Two government agency employees that are part of the AI team at their organization in the Americas that has 12,000 employees. These interviewees were responsible for reviewing approximately 90,000 files weekly.

Prior to using Microsoft Azure OpenAI Service, the interviewees from government organizations attempted to service high volumes of constituents and documents with internally developed or human-powered solutions. As a result, technology gaps led to operational inefficiencies and inconsistent data analysis that impacted the internal employees’ abilities to service the public effectively.

Interviewees noted Microsoft Azure OpenAI Service enabled their public sector organizations to develop custom AI applications that use Microsoft’s underlying AI infrastructure. This strategy allowed AI to impact as many applications, business processes, and employees as possible while still being accessible and manageable for AI developers. The interviewees’ government organizations used Azure OpenAI Service to build AI solutions to meet their specific needs:

• The AI lead supporting government agencies described an AI model that was trained on region-specific legal judgements so that it could read and annotate those judgements in seconds.
• This interviewee also described how AI was used to scrape schemes from their website and make them available to their constituents through a bot.   
• The two government agency employees utilized AI to improve existing text and inform future writings with more detailed information.
• The same interviewees at the government agency also used AI to extract important information from files and petitions in a structured way.

*For the composite organization presented in the full TEI study

INVESTMENT DRIVERS

The interviewees’ organizations adopted Azure OpenAI Service to overcome various challenges in their legacy environments, including:

• Resource constraints. One government employee indicated that before Azure OpenAI Service, lawyers were manually reviewing high volumes of long, complicated documents, which resulted in a time-sink for these highly valuable resources. The interviewee stated: “It doesn’t make sense to have our lawyers with expensive rates do this type of task. We want them to do more complex analysis.” For the AI lead supporting government agencies in APAC, the delayed delivery times resulted in delayed legal processes for their constituents.
• Inconsistencies in data analysis. Manual processes often resulted in data analysis errors, leading to confusion and productivity losses. One interviewee at the government agency indicated that two different employees could come up with different analyses when reviewing similar cases with their prior review methods.
• Technology infrastructure gaps. Without proper technology to support complex government environments, the interviewees’ organizations often resorted to applying more internal employees to the problem, which further compounded operational inefficiencies and data inconsistencies. The AI lead supporting government agencies described how the technology gaps directly impacted constituents’ abilities to understand schemes and receive legal support: “[Our country’s judgement] is notoriously long and hard to understand. You would need an army of lawyers to understand. However, 99% of [our constituents] cannot afford a lawyer.”

key results

The results of the investment for the interviewees’ organizations include:

Optimized spend. Interviewees at government organizations cited examples of how their use of Microsoft Azure OpenAI Service proved critical in reducing operational expenses.

Azure OpenAI Service enabled the interviewees’ organizations to redistribute spending to focus on higher-value activities. The AI lead supporting government agencies noted: “Our government has the mandate to improve access to [public service]. Right now, we have people on the ground whose job it is to disseminate information about specific [government programs]. These people are doing this work voluntarily, not dedicated to this role. By having chatbots on people’s phones, we probably only need 10% of the effort, which means these people can return to their main responsibilities.”

Increased operational efficiencies. Interviewees noted that Azure OpenAI Service expedited internal processes and automated certain tasks to provide operational efficiencies in a few different areas including providing support to the general public and constituents.

• The AI lead supporting government agencies indicated an AI use case that enabled farmers in an agriculture program to use chatbots to take photos of their crops and receive recommendations and suggestions. The interviewee noted, “[This specific example signifies a] popular use case across many domains where frontline workers are empowered with some sort of pocket encyclopedia for a specific domain.”
• One interviewee from the government organization in the Americas shared that their employees received suggestions from the larger internal knowledge base by asking questions through Azure OpenAI Service. The government employee stated, “The value here is time savings and good answers.”
• The interviewees at the government organization in the Americas utilized OpenAI to create a text editor. This AI assistant helped their government employees write new documents or improve the writing of an existing document.

Improved service delivery. Interviewees noted that ultimately, refocusing organizational efforts to reduce administrative burdens and providing operational efficiencies resulted in enhanced service delivery in terms of both timelines and quality.

• The AI lead supporting government agencies noted how AI improved accessibility of the agencies’ services, stating, “You can talk to our chatbot in any language — both text and voice — give your description of the situation, and it would generate a response that is applicable to you.”
• An interviewee at a government agency cited massive time savings in one of their AI use cases, “Depending on the case, we are saving as much as 80% of our time by having AOAI review our documents.” The outcome of this time savings was in the ability to expedite the legal process for constituents. The AI lead supporting government agencies described a use case: “AI is making information easily accessible to people without having to send a physical agent on the ground for different domains. We are talking about access to legal services, primary healthcare, and education.”
• The AI lead supporting government agencies tied more efficient service delivery to better constituent and public satisfaction by citing, again, the use case where AI provided domain-specific knowledge to people in lieu of physical agents. While this use case provided time savings, it also enabled a type of service for constituents that was not previously possible at the organization. The AI lead stated, “More than improving productivity, it’s filling a gap with something where nothing existing prior.”
• An interviewee at a government agency established that AI automated certain tasks that allowed lawyers to focus on more complex matters. Overall, this resulted in better decision-making across their government organization, which directly impacted client satisfaction by avoiding unnecessary litigation and prolonged case times. One interviewee at a government agency elaborated: “Think about a citizen with rights. If they need to spend hours, days, months, years to fight for a right they have, that hurts them. That hurts us as a county. If we can avoid cases that spend years fighting for a citizen’s rights, that would be a big societal win.”

Appendix A: Endnotes