Nowadays, governments around the world are facing substantial challenges. Budget constraints arising from macroeconomic shocks and geopolitical conditions demand cost efficiency from the government back office (finance, HR, procurement, etc.), while citizens are raising their expectations for highly agile, trustworthy, and effective services.
This is where SAP Business Artificial Intelligence (SAP Business AI) steps in. It encompasses all the latest AI technologies, such as Machine Learning (ML), Generative AI, AI Copilots and most recently AI agents, ensuring the public sector can meet these ever-growing demands.
SAP Business AI completely redefines the way governments are operating their business system – it offers one single SAP Joule copilot that serves as an embedded digital assistant. In the daily work of public sector employees, Joule navigates their system landscapes, answers questions, prepares data call outs, and executes transactions. Joule Copilot is deeply embedded in the context of your data and works across all departments and applications.
This year, SAP Joule is evolving from a simple AI copilot to being powered by AI Agents. It will continue to utilize ML and Generative AI skills but at the same time it has advanced its intelligence at an unprecedented scale. Embedded in Joule and the broader SAP Business AI portfolio, AI agents offer a new level of task execution autonomy, helping governments optimize their resources across their systems in the most efficient way.
Agents don’t just perform tasks faster—they possess near-human intelligence, allowing them to boost productivity for government services, offer personalization, and provide 24/7 resolution support.
In this blog, we will delve into what SAP Agentic AI means for the public sector and share some concrete use case examples.
Agentic AI for Governments – The Edge over GenAI
So, how Agentic AI works? It begins with Large Language Models (LLMs). Trained on the vast amount of data these models are able to provide relevant suggestions to a broad set of questions – we call it Generative AI (GenAI). AI Agents leverage more advanced LLMs called Reasoning Language Models (RLM), that are enhanced with the ability to reason. This gives them an opportunity to not only answer questions but solve complex tasks. When handling a problem, an agent first breaks it down into smaller, more manageable chunks, solves each one, and finally combines the results into a complete and accurate solution.
Unlike a GenAI-powered copilot, which needs a human to tell it what to do before acting, agentic AI can make its own decisions and act proactively. This thinking capability enables four main core abilities: planning, tool usage, reflection, and collaboration.
Let’s look at these properties by picking up an example of a typical task public sector authorities are regularly dealing with – answering citizen queries about their citizenship application status. We will see the extra benefits agents can provide when compared to GenAI assistants.
How a Gen-AI Copilot would handle a citizenship process
How an AI Agent would handle a citizenship process
PlanningGenAI Copilot don´t plan resolution steps to answer their request; it only follow the instructions received from the user. Its ability to assist would be limited to answering direct questions asked by the applicant who is executing the plan and actions by himself. These questions would be usually limited to understanding passport renewal application process.
The AI Agent plans a series of steps with a concrete plan to achieve by itself the given task from the request. It therefore goes beyond just answering questions about the process.
While it is true that before applying citizens are learning about their next steps, when they are waiting for a decision, they typically reach out with specific concerns that they want to resolve – for example “My process is taking longer than the time mentioned in the service description – can you address this?”. Unlike GenAI Copilot, AI Agents can assist in solving highly complex issues.
Tool usage
The GenAI Copilot can access information through tools but cannot execute actions with them. It does not perform tasks in the system; it only provides information based on what it sees. If an applicant is complaining about the process delay, GenAI Copilot would likely be helpful in double checking the process policy and confirming the waiting time has exceeded the regulated one. It can also be helpful in listing the contacts to reach out to. But it won’t progress further than this.
An agent has a unique ability to reach out to connected tools to carry out the resolution plan. It can choose the tools on its own, use them to consult information, but also to execute actions directly in the system. These tools include databases, software, internet search, a calculator, and more. This helps it to not only advise on the best next actions but execute them.
In our example, by using APIs and database access, an agent can fetch the right application data, create a dispute ticket, describe an issue, assign it to the right human responsible, and send a confirmation to the applicant.
Collaboration
The GenAI Copilot does not consult other collaborators in real time while executing its tasks. It follows the instructions it received and responds to requests based on its own knowledge.
Agents are enjoying teamwork. The AI Agents can call for help when they need it. They can ask a human during their task to supervise some of their actions or request clarifications.
They can also ask for a helping hand from other specialized AI agents to carry out specific tasks that require specialized knowledge. By collaborating together, multi-agent systems are even more effective at solving problems.
In our case, this collaboration ability is crucial. Only human processors responsible for a citizenship case can decide on the final resolution decision. So, our agent will involve them in the review of all critical steps made in the resolution process and will ask them to define the final decisions.
When specific rules apply, it will ask a Rule-specialized Agent to provide it with the right guidance and knowledge to tackle its task.
Reflection
The GenAI Copilot mainly produces text output without genuine self-evaluation. It does not review the results it produces. The quality of its output can improve only if the user refines their prompt (prompt engineering) or if the underlying LLMs receive additional training.
The AI Agent will do its best to achieve its goal. As soon as it encounters an obstacle or is at the end of a processus, it will self-evaluate to see if it meets the desired output, then corrects itself, and explores multiple paths until the goal is reached. It learns by itself from his experience to get better next time.
Returning to our citizen process case, it means that in case an agent won’t be able to find the applicant data in one database, it will check another one. It will learn next time where it can find all the necessary information the quickest.
For public sector agencies the shift from GenAI towards AI agents translates in two major benefits:
Higher efficiency and cost reduction: unlike GenAI that automates single content related tasks, agents can take care of complete business processes with more proactiveness and intelligence speeding up large government operations.Higher effectiveness and autonomy: applied in critical areas, advanced agents empower governments to act continuously – even outside of normal hours and will with limited human supervision – extending citizen support and increasing preparedness in disaster and response management.
Those benefits will span across the whole organization—from front‑office touchpoints to back‑office department— bringing public sector entities to their next level.
Agentic AI Drives Efficiency in the Government Back Office
Like any other organization, a public sector agency needs to be efficient in running its operational processes. Those include procuring goods and services, hiring and training employees, and of course, planning and managing public finances. Back-office tasks involve a lot of repetitive tasks, such as record-keeping, document management, transaction processing, and much more. They often rely on manual workflows that can be time-consuming, error-prone, and non-transparent.
But the biggest challenge the governments are facing in their back-office operations is not being able to scale during periods of unprecedented workload.
Starting from the end of 2024, SAP Joule has been successfully helping governments to leverage GenAI to reduce the operational burden and improve resilience.
Applied right at the center of government finance, S/4HANA Public Sector Management, Joule is already assisting with many content-related tasks such as answering questions about key financial processes in SAP systems, understanding main financial objects and entities and navigating users through SAP applications.
All of it leads to 95% improvement in the speed of learning, search and navigation – from 5 minutes usually spent on these tasks to just 30 seconds!
Learn more about Joule for SAP Public Sector Management
Introducing Agentic AI with Joule, let’s explore how innovation can further drive operational excellence and in the back office.
Agents are now able to support tasks of a much higher complexity. In the finance domain, this would mean not only content support and navigation, but the execution of financial transactions themselves. An example of a financial transaction is a budget transfer. Let’s explore the ways an agent will be able to support.
Let’s imagine a team manager is considering an extensive overseas business trip. She has a cost estimate but is not sure if she needs special approval from her department director. With the help of document grounding, a budget agent can analyze the travel policy of her agency and confirm her estimate is within the threshold, which means she can go ahead with flight and hotel booking without any extra approvals.
She is then able to reserve some budget. As she is also the cost center responsible for her team, a budget agent is permitting her to look at the available funds. It pulls the overall cost center budget for the year and informs her about the remaining funds. She quickly notices that her business trip estimate is slightly exceeding the travel funds available. She asks the agent to reallocate some funds from the Team Activities budget bucket to the Travel bucket. An SAP agent is then able to confirm the transaction details, create a budget document, and trigger the transfer.
SAP Budget Agent visionary demo screen
SAP agents are operating with the permissions of a user and within the organizational context and can assist with a broader set of tasks then a GenAI Copilot: they are not only communicating in natural language but are also able to perform calculations and work with statistical insights.
Due their extended intelligence and end-to-end support of public finance processes that include many other typically executed tasks like booking commitments or requesting supplements, budget agents will be able to bring an extra efficiency boost of around 30%.
While it is true that SAP public finance agents will be able to take care of many tasks on behalf of government users, humans will always remain in control. To ensure this, each agent is first performing in a draft mode, so that public sector specialists can review the resolution steps and then approve or modify the final decisions.
As noted earlier, agents are particularly effective in managing unprecedented workloads — scenarios in which human capacity alone would be insufficient. Let’s look at one very recent customer example using near Agentic AI technology —an advanced, GenAI‑powered, context‑aware automation system— that demonstrate the potential impact that Agentic AI can have in back-office processes.
The city of Antibes in south of France is not only efficient in its public finance but also is making sure its spending is contributing to making Antibes sustainable and eco-friendly.
Antibes CIO, Patrick Duverger, has led the city to winning multiple innovation awards, including the Territoria Gold Award 2023 by the French Government and SAP Innovation Award 2025
Since the new green law for municipalities came into force, Antibes were looking into “greening” its budget. This means the city is expected to quantify the environmental impact of their annual budget by tagging expenses to 6 EU Taxonomy objectives and 17 UN Sustainable Development Goals (SDGs). In practice, it means attributing 6,000 budget expenses to one or many green objectives – which results in making over 138,000 decisions! Clearly this is beyond a human capacity to manage.
Leveraging Business AI with its advanced natural language processing, SAP was able to help.
Based on the descriptions of spend categories and those of the green objectives, AI was able to map all 6,000 budget categories to the relevant sustainable development objective in just about 15-20 minutes!
This allowed the city to “green” their finances by providing a view to their inhabitants on how much of Antibes’ annual budget was spent on each of the EU and the UN goals. Thanks to SAP Business, residents of Antibes can clearly see how the city is contributing to a better environment and how public funds are being used — fostering greater transparency and trust.
Agentic AI Transforms Citizen Service Delivery
In addition to handling administrative routing tasks in the back office, SAP Agentic AI will also change the way governments are serving their constituents.
Today citizens expect more straightforward, personalized, and agile services, and it is crucial for public sector agencies to meet these expectations – at a local, regional or federal level, they are eager to change the way they communicate and interact with the service recipients.
The overwhelming number of inquiries and human resource constraints is a common challenge to overcome and here SAP Business AI brings an immense value.
A notable example from SAP Business AI impact is the German Federal Foreign Office, which is using SAP’s Business Technology Platform (SAP BTP) to streamline their services.
The Office manages Germany’s foreign relations with about 230 global missions worldwide. They are dealing with a high volume of inquiries globally and use multiple communication channels.
Using SAP BTP powered by natural language processing, as well as GenAI and Machine Learning capabilities, they have developed a centralized omnichannel communication processing system that integrates mail, tax, phone, portal and other correspondence modes in one single workflow. Using a ticketing system powered by AI, they were able to automatically classify and assign inquiries to the right processor in their organization.
Furthermore, SAP Business AI has been recommending the best resolution to each of the requests, relying on already resolved inquiries.
Since the project’s launch in 2021 through the completion of its phased rollout in 2023, the German Federal Foreign Office successfully leveraged AI to process more than 2 million messages and over 750,000 inquiries. Today, 50% of all incoming cases are resolved using AI-assisted template responses, eliminating the need for human intervention in those interactions.
For their AI project with SAP, the German Federal Foreign Office received first prize in the “Services Supernova” category at the 2024 SAP Innovation Awards
As a result, Federal Office employees managed to avoid inconsistent answers to similar questions, assign clear processing and status tracking responsibilities to the teams and, most importantly increase service speed – 77% of inquiries are now answered within the same day!
Let’s now illustrate how this can translate to Agentic AI and transform citizen service delivery even better – based on one of the most complex service examples.
Consider a service officer of a local city council who handles high volumes of complex building permit applications. As legal requirements differ significantly depending on the object that is being built – requirements for a house construction are different from those for a swimming pool, for instance – ensuring regulatory compliance while managing strict service deadlines is challenging.
In addition to this, paper-based applications are frequently misplaced and subject to delays, thereby exacerbating the workload of service officers and resulting in applicant dissatisfaction.
A multi-agentic AI workflow from SAP would be ideal to address the situation.
Here’s how it can work. A builder submits a digital request for a new house construction, an AI-driven automation generates an application case. A multi-agentic AI team, consisting of four specialized agents is then stepping in.
A document agent ensures application completeness and checks if all required documents are in place and all required information in entered. A permit rules agent extracts decision-making criteria from the legal regulations and knows which criteria an application should fulfill to be granted an approval. It then validates the submitted evidence against the requirements and advises to the service officer if a positive decision should be granted. A clock agent helps to ensure all processing steps are done within the maximum service delivery time and helps to schedule in-person inspections with the builder. A service agent is there to draft personalized emails to the builder, confirming application entrance, payments, and scheduled inspections.
SAP Application Approval agents visionary demo screen
By leveraging Agentic AI, public sector officers can simplify the processing of complex services, such as building permit applications, expediting turnaround times while enhancing regulatory compliance with more accurate decision-making and applicant satisfaction.
New Horizons in Government Transformation with AI Agents
As we have seen AI Agents hold the keys to have a concrete application and real impact whether it is in front office or back office. Let us summarize the broader value public sector organizations will be able to seize from Agentic AI.
In the back office that unites key government operations such as finance, HR, and procurement AI agents are boosting productivity by up to 30%. They act as true teammates and make work more efficient, delivering more with fewer resources and at lower cost:
AI agents boost civil servants’ productivity by executing repetitive manual tasks using specialized skills, expertise and knowledge – this reduces and prevent errors and allow to process high workloads.They can operate autonomously and across multiple departments, involving humans only at critical decision-making points – freeing up their time to focus on making those strategic decisions effectively.Agentic AI helps new government hires to get a hang of complex back-office processes faster and run them with ease.
In the context of citizen services, AI agents provide scalability in times of high demand and enhance citizens satisfaction and trust in government:
Agentic AI is redefining the citizen experience by providing 24/7 support to resolve less critical issues, reduce friction points, and significantly shorten processing times for more complex cases. It also increases personalization by tailoring the content and making decisions based on the most relevant data.AI agents provide vital support during periods of exceptionally high demand. When unemployment rises or other economic shocks occur, and the public sector faces overwhelming volumes of service requests, AI agents help manage workloads beyond human capacity and allow civil servant to stay available in every scenario to answer special demands requiring special attention and compassion that only a human can do.On top, civil servants can free prioritize on citizen transparency, leveraging agents to help with public reporting and hence increase constituents trust.
We recognize that despite the power of agentic AI, humans remain accountable for public sector decisions and use of this technology. Accordingly, SAP AI agents are designed under the SAP Global AI Ethics Policy with risk-calibrated autonomy and with great traceability to ensure that humans stay in control to review and approve work performed by agents.
With this considered, Agents are a true revolution in the AI space – thanks to their advanced intelligence, governments have a new window of opportunity to simplify the way they work and to deliver next level impact in a positive and responsible way to their citizens.
Nowadays, governments around the world are facing substantial challenges. Budget constraints arising from macroeconomic shocks and geopolitical conditions demand cost efficiency from the government back office (finance, HR, procurement, etc.), while citizens are raising their expectations for highly agile, trustworthy, and effective services.This is where SAP Business Artificial Intelligence (SAP Business AI) steps in. It encompasses all the latest AI technologies, such as Machine Learning (ML), Generative AI, AI Copilots and most recently AI agents, ensuring the public sector can meet these ever-growing demands.SAP Business AI completely redefines the way governments are operating their business system – it offers one single SAP Joule copilot that serves as an embedded digital assistant. In the daily work of public sector employees, Joule navigates their system landscapes, answers questions, prepares data call outs, and executes transactions. Joule Copilot is deeply embedded in the context of your data and works across all departments and applications.This year, SAP Joule is evolving from a simple AI copilot to being powered by AI Agents. It will continue to utilize ML and Generative AI skills but at the same time it has advanced its intelligence at an unprecedented scale. Embedded in Joule and the broader SAP Business AI portfolio, AI agents offer a new level of task execution autonomy, helping governments optimize their resources across their systems in the most efficient way.Agents don’t just perform tasks faster—they possess near-human intelligence, allowing them to boost productivity for government services, offer personalization, and provide 24/7 resolution support.In this blog, we will delve into what SAP Agentic AI means for the public sector and share some concrete use case examples. Agentic AI for Governments – The Edge over GenAISo, how Agentic AI works? It begins with Large Language Models (LLMs). Trained on the vast amount of data these models are able to provide relevant suggestions to a broad set of questions – we call it Generative AI (GenAI). AI Agents leverage more advanced LLMs called Reasoning Language Models (RLM), that are enhanced with the ability to reason. This gives them an opportunity to not only answer questions but solve complex tasks. When handling a problem, an agent first breaks it down into smaller, more manageable chunks, solves each one, and finally combines the results into a complete and accurate solution.Unlike a GenAI-powered copilot, which needs a human to tell it what to do before acting, agentic AI can make its own decisions and act proactively. This thinking capability enables four main core abilities: planning, tool usage, reflection, and collaboration.Let’s look at these properties by picking up an example of a typical task public sector authorities are regularly dealing with – answering citizen queries about their citizenship application status. We will see the extra benefits agents can provide when compared to GenAI assistants. How a Gen-AI Copilot would handle a citizenship process How an AI Agent would handle a citizenship processPlanningGenAI Copilot don´t plan resolution steps to answer their request; it only follow the instructions received from the user. Its ability to assist would be limited to answering direct questions asked by the applicant who is executing the plan and actions by himself. These questions would be usually limited to understanding passport renewal application process.The AI Agent plans a series of steps with a concrete plan to achieve by itself the given task from the request. It therefore goes beyond just answering questions about the process.While it is true that before applying citizens are learning about their next steps, when they are waiting for a decision, they typically reach out with specific concerns that they want to resolve – for example “My process is taking longer than the time mentioned in the service description – can you address this?”. Unlike GenAI Copilot, AI Agents can assist in solving highly complex issues.Tool usageThe GenAI Copilot can access information through tools but cannot execute actions with them. It does not perform tasks in the system; it only provides information based on what it sees. If an applicant is complaining about the process delay, GenAI Copilot would likely be helpful in double checking the process policy and confirming the waiting time has exceeded the regulated one. It can also be helpful in listing the contacts to reach out to. But it won’t progress further than this.An agent has a unique ability to reach out to connected tools to carry out the resolution plan. It can choose the tools on its own, use them to consult information, but also to execute actions directly in the system. These tools include databases, software, internet search, a calculator, and more. This helps it to not only advise on the best next actions but execute them. In our example, by using APIs and database access, an agent can fetch the right application data, create a dispute ticket, describe an issue, assign it to the right human responsible, and send a confirmation to the applicant.CollaborationThe GenAI Copilot does not consult other collaborators in real time while executing its tasks. It follows the instructions it received and responds to requests based on its own knowledge.Agents are enjoying teamwork. The AI Agents can call for help when they need it. They can ask a human during their task to supervise some of their actions or request clarifications.They can also ask for a helping hand from other specialized AI agents to carry out specific tasks that require specialized knowledge. By collaborating together, multi-agent systems are even more effective at solving problems.In our case, this collaboration ability is crucial. Only human processors responsible for a citizenship case can decide on the final resolution decision. So, our agent will involve them in the review of all critical steps made in the resolution process and will ask them to define the final decisions.When specific rules apply, it will ask a Rule-specialized Agent to provide it with the right guidance and knowledge to tackle its task.ReflectionThe GenAI Copilot mainly produces text output without genuine self-evaluation. It does not review the results it produces. The quality of its output can improve only if the user refines their prompt (prompt engineering) or if the underlying LLMs receive additional training.The AI Agent will do its best to achieve its goal. As soon as it encounters an obstacle or is at the end of a processus, it will self-evaluate to see if it meets the desired output, then corrects itself, and explores multiple paths until the goal is reached. It learns by itself from his experience to get better next time.Returning to our citizen process case, it means that in case an agent won’t be able to find the applicant data in one database, it will check another one. It will learn next time where it can find all the necessary information the quickest.For public sector agencies the shift from GenAI towards AI agents translates in two major benefits:Higher efficiency and cost reduction: unlike GenAI that automates single content related tasks, agents can take care of complete business processes with more proactiveness and intelligence speeding up large government operations.Higher effectiveness and autonomy: applied in critical areas, advanced agents empower governments to act continuously – even outside of normal hours and will with limited human supervision – extending citizen support and increasing preparedness in disaster and response management.Those benefits will span across the whole organization—from front‑office touchpoints to back‑office department— bringing public sector entities to their next level. Agentic AI Drives Efficiency in the Government Back Office Like any other organization, a public sector agency needs to be efficient in running its operational processes. Those include procuring goods and services, hiring and training employees, and of course, planning and managing public finances. Back-office tasks involve a lot of repetitive tasks, such as record-keeping, document management, transaction processing, and much more. They often rely on manual workflows that can be time-consuming, error-prone, and non-transparent.But the biggest challenge the governments are facing in their back-office operations is not being able to scale during periods of unprecedented workload. Starting from the end of 2024, SAP Joule has been successfully helping governments to leverage GenAI to reduce the operational burden and improve resilience.Applied right at the center of government finance, S/4HANA Public Sector Management, Joule is already assisting with many content-related tasks such as answering questions about key financial processes in SAP systems, understanding main financial objects and entities and navigating users through SAP applications.All of it leads to 95% improvement in the speed of learning, search and navigation – from 5 minutes usually spent on these tasks to just 30 seconds!Learn more about Joule for SAP Public Sector Management Introducing Agentic AI with Joule, let’s explore how innovation can further drive operational excellence and in the back office.Agents are now able to support tasks of a much higher complexity. In the finance domain, this would mean not only content support and navigation, but the execution of financial transactions themselves. An example of a financial transaction is a budget transfer. Let’s explore the ways an agent will be able to support.Let’s imagine a team manager is considering an extensive overseas business trip. She has a cost estimate but is not sure if she needs special approval from her department director. With the help of document grounding, a budget agent can analyze the travel policy of her agency and confirm her estimate is within the threshold, which means she can go ahead with flight and hotel booking without any extra approvals.She is then able to reserve some budget. As she is also the cost center responsible for her team, a budget agent is permitting her to look at the available funds. It pulls the overall cost center budget for the year and informs her about the remaining funds. She quickly notices that her business trip estimate is slightly exceeding the travel funds available. She asks the agent to reallocate some funds from the Team Activities budget bucket to the Travel bucket. An SAP agent is then able to confirm the transaction details, create a budget document, and trigger the transfer. SAP Budget Agent visionary demo screen SAP agents are operating with the permissions of a user and within the organizational context and can assist with a broader set of tasks then a GenAI Copilot: they are not only communicating in natural language but are also able to perform calculations and work with statistical insights.Due their extended intelligence and end-to-end support of public finance processes that include many other typically executed tasks like booking commitments or requesting supplements, budget agents will be able to bring an extra efficiency boost of around 30%.While it is true that SAP public finance agents will be able to take care of many tasks on behalf of government users, humans will always remain in control. To ensure this, each agent is first performing in a draft mode, so that public sector specialists can review the resolution steps and then approve or modify the final decisions.As noted earlier, agents are particularly effective in managing unprecedented workloads — scenarios in which human capacity alone would be insufficient. Let’s look at one very recent customer example using near Agentic AI technology —an advanced, GenAI‑powered, context‑aware automation system— that demonstrate the potential impact that Agentic AI can have in back-office processes.The city of Antibes in south of France is not only efficient in its public finance but also is making sure its spending is contributing to making Antibes sustainable and eco-friendly.Antibes CIO, Patrick Duverger, has led the city to winning multiple innovation awards, including the Territoria Gold Award 2023 by the French Government and SAP Innovation Award 2025Since the new green law for municipalities came into force, Antibes were looking into “greening” its budget. This means the city is expected to quantify the environmental impact of their annual budget by tagging expenses to 6 EU Taxonomy objectives and 17 UN Sustainable Development Goals (SDGs). In practice, it means attributing 6,000 budget expenses to one or many green objectives – which results in making over 138,000 decisions! Clearly this is beyond a human capacity to manage.Leveraging Business AI with its advanced natural language processing, SAP was able to help.Based on the descriptions of spend categories and those of the green objectives, AI was able to map all 6,000 budget categories to the relevant sustainable development objective in just about 15-20 minutes!This allowed the city to “green” their finances by providing a view to their inhabitants on how much of Antibes’ annual budget was spent on each of the EU and the UN goals. Thanks to SAP Business, residents of Antibes can clearly see how the city is contributing to a better environment and how public funds are being used — fostering greater transparency and trust. Agentic AI Transforms Citizen Service DeliveryIn addition to handling administrative routing tasks in the back office, SAP Agentic AI will also change the way governments are serving their constituents.Today citizens expect more straightforward, personalized, and agile services, and it is crucial for public sector agencies to meet these expectations – at a local, regional or federal level, they are eager to change the way they communicate and interact with the service recipients.The overwhelming number of inquiries and human resource constraints is a common challenge to overcome and here SAP Business AI brings an immense value.A notable example from SAP Business AI impact is the German Federal Foreign Office, which is using SAP’s Business Technology Platform (SAP BTP) to streamline their services.The Office manages Germany’s foreign relations with about 230 global missions worldwide. They are dealing with a high volume of inquiries globally and use multiple communication channels.Using SAP BTP powered by natural language processing, as well as GenAI and Machine Learning capabilities, they have developed a centralized omnichannel communication processing system that integrates mail, tax, phone, portal and other correspondence modes in one single workflow. Using a ticketing system powered by AI, they were able to automatically classify and assign inquiries to the right processor in their organization.Furthermore, SAP Business AI has been recommending the best resolution to each of the requests, relying on already resolved inquiries.Since the project’s launch in 2021 through the completion of its phased rollout in 2023, the German Federal Foreign Office successfully leveraged AI to process more than 2 million messages and over 750,000 inquiries. Today, 50% of all incoming cases are resolved using AI-assisted template responses, eliminating the need for human intervention in those interactions.For their AI project with SAP, the German Federal Foreign Office received first prize in the “Services Supernova” category at the 2024 SAP Innovation AwardsAs a result, Federal Office employees managed to avoid inconsistent answers to similar questions, assign clear processing and status tracking responsibilities to the teams and, most importantly increase service speed – 77% of inquiries are now answered within the same day!Let’s now illustrate how this can translate to Agentic AI and transform citizen service delivery even better – based on one of the most complex service examples.Consider a service officer of a local city council who handles high volumes of complex building permit applications. As legal requirements differ significantly depending on the object that is being built – requirements for a house construction are different from those for a swimming pool, for instance – ensuring regulatory compliance while managing strict service deadlines is challenging.In addition to this, paper-based applications are frequently misplaced and subject to delays, thereby exacerbating the workload of service officers and resulting in applicant dissatisfaction.A multi-agentic AI workflow from SAP would be ideal to address the situation. Here’s how it can work. A builder submits a digital request for a new house construction, an AI-driven automation generates an application case. A multi-agentic AI team, consisting of four specialized agents is then stepping in.A document agent ensures application completeness and checks if all required documents are in place and all required information in entered. A permit rules agent extracts decision-making criteria from the legal regulations and knows which criteria an application should fulfill to be granted an approval. It then validates the submitted evidence against the requirements and advises to the service officer if a positive decision should be granted. A clock agent helps to ensure all processing steps are done within the maximum service delivery time and helps to schedule in-person inspections with the builder. A service agent is there to draft personalized emails to the builder, confirming application entrance, payments, and scheduled inspections.SAP Application Approval agents visionary demo screenBy leveraging Agentic AI, public sector officers can simplify the processing of complex services, such as building permit applications, expediting turnaround times while enhancing regulatory compliance with more accurate decision-making and applicant satisfaction. New Horizons in Government Transformation with AI AgentsAs we have seen AI Agents hold the keys to have a concrete application and real impact whether it is in front office or back office. Let us summarize the broader value public sector organizations will be able to seize from Agentic AI.In the back office that unites key government operations such as finance, HR, and procurement AI agents are boosting productivity by up to 30%. They act as true teammates and make work more efficient, delivering more with fewer resources and at lower cost:AI agents boost civil servants’ productivity by executing repetitive manual tasks using specialized skills, expertise and knowledge – this reduces and prevent errors and allow to process high workloads.They can operate autonomously and across multiple departments, involving humans only at critical decision-making points – freeing up their time to focus on making those strategic decisions effectively.Agentic AI helps new government hires to get a hang of complex back-office processes faster and run them with ease.In the context of citizen services, AI agents provide scalability in times of high demand and enhance citizens satisfaction and trust in government:Agentic AI is redefining the citizen experience by providing 24/7 support to resolve less critical issues, reduce friction points, and significantly shorten processing times for more complex cases. It also increases personalization by tailoring the content and making decisions based on the most relevant data.AI agents provide vital support during periods of exceptionally high demand. When unemployment rises or other economic shocks occur, and the public sector faces overwhelming volumes of service requests, AI agents help manage workloads beyond human capacity and allow civil servant to stay available in every scenario to answer special demands requiring special attention and compassion that only a human can do.On top, civil servants can free prioritize on citizen transparency, leveraging agents to help with public reporting and hence increase constituents trust.We recognize that despite the power of agentic AI, humans remain accountable for public sector decisions and use of this technology. Accordingly, SAP AI agents are designed under the SAP Global AI Ethics Policy with risk-calibrated autonomy and with great traceability to ensure that humans stay in control to review and approve work performed by agents.With this considered, Agents are a true revolution in the AI space – thanks to their advanced intelligence, governments have a new window of opportunity to simplify the way they work and to deliver next level impact in a positive and responsible way to their citizens. Read More Technology Blog Posts by SAP articles
#SAP
#SAPTechnologyblog
