ERP and CRM Transformation with Agentic AI: A Global and Turkish Perspective

ERP and CRM Transformation with Agentic AI: A Global and Turkish Perspective

In the corporate world, the use of artificial intelligence is evolving beyond fixed reporting and predefined processes into a more dynamic and autonomous structure. Agentic AI is the newest instrument businesses have in this transformation. According to Gartner’s forecast, by 2028, 15% of day-to-day operational decisions will be made autonomously by agentic AI (this rate was 0% in 2024), and one-third of enterprise software will include agentic AI. So what exactly is agentic AI? Agentic AI is not a simple automation step or an ordinary generative AI application. Agentic AI systems can carry out complex tasks on their own without requiring continuous human approval, optimize processes, and proactively identify opportunities or risks. This makes it possible to respond quickly to changing market conditions, improve decision-making processes, and redirect human resources toward higher value-added work.

The shift from static reporting to dynamic decision support is one of the core promises of agentic AI. In traditional ERP and CRM systems, executives typically receive static reports based on historical data and support their decisions with these reports. Agentic AI, however, creates dynamic decision-support systems through real-time and context-aware analyses. For example, SAP’s next-generation AI assistant Joule aims to provide deep analytics and reporting by blending both internal enterprise data and external intelligence on a single platform; moreover, users can ask the system complex questions in natural language and receive answers containing strategic insights. This kind of proactive decision support enables executives to use data not only in retrospective reports but also as an instantaneous and forward-looking guide. As a result, agentic AI enables the transition from static business intelligence to a dynamic and interactive governance model by offering enterprises a continuously learning, adaptive, and human-assisting digital workforce.

Agentic AI Integration in ERP and CRM Processes

Enterprise Resource Planning (ERP) and Customer Relationship Management (CRM) processes contain an organization’s most critical data and workflows. When agentic AI is integrated into these areas, it fundamentally changes how data and workloads are managed. In particular, global software providers such as SAP and Salesforce offer leading examples that accelerate transformation by strengthening ERP and CRM platforms with agentic AI. In this section, we will examine SAP’s Joule AI and Business Network initiatives as well as Salesforce’s Agentforce 360 platform. We will analyze how these platforms affect data integration and workload automation, and how they enable the shift from static systems to a dynamic structure. Additionally, current examples and gains related to this transformation—both globally and in Türkiye—will be addressed.

Enterprise Transformation with SAP Joule AI and Business Network

SAP is opening a new chapter in ERP by combining nearly 50 years of enterprise process expertise with artificial intelligence. SAP Joule AI is the suite of AI copilot capabilities and autonomous agents integrated into SAP’s cloud-based ERP and business applications. Joule sits at the heart of SAP’s “AI-first” strategy, and its goal is to provide proactive support to business users across every department. Joule Agents are autonomous AI agents embedded within SAP systems to deliver decision support and task automation. These agents can manage non-deterministic workflows that cannot be constrained by predefined rule sets; to achieve a goal, they dynamically select the tools and other sub-agents they need and make recommendations by evaluating the results. Supported by SAP’s comprehensive Knowledge Graph and Business Data Cloud infrastructure, these agents can take reliable and context-appropriate actions because they deeply understand enterprise data and processes.

To bring the Joule AI vision to life, SAP is developing intelligent agents tailored to different ERP functions. At the end of 2025, SAP announced that it had restructured the Joule platform from a reactive assistant approach into a network of specialized agents that enables end-to-end automation of business processes. In this direction, SAP introduced 15 new Joule agents in critical areas such as supply chain, finance, human resources, and customer experience. For example, the Production Planning and Operations Agent automatically performs the necessary checks before production orders are released, detects potential blockers, and offers solution suggestions to production managers. The Change Management Agent analyzes change requests arising in the product lifecycle and recommends the most suitable action plan to complete the project. The Supplier Onboarding Agent, which is directly related to SAP’s Business Network, automates the end-to-end process of bringing suppliers onto the SAP Business Network. This AI agent reviews supplier information, manages invitation processes, tracks onboarding steps, and notifies relevant teams in case of issues. SAP emphasizes that thanks to automation, this process reduces the operational workload of procurement teams, allowing them to focus on more strategic work.

Joule AI has also begun to show tangible contributions in customer service and CRM. For example, the global technology and engineering company Bosch reports that it achieved striking improvements in customer support processes by integrating an AI-supported case classification agent within SAP Service Cloud. Bosch’s Director of Digital Customer Experience, Florian Haustein, states that they replaced hundreds of manually defined routing rules with a single AI-enabled prompt, increasing the accuracy of directing support requests to the right units, reducing manual effort, and accelerating resolution times. This example provides compelling proof of how agentic AI-supported ERP/CRM integration can generate faster and more accurate outcomes with less human intervention in the real world.

Transformation in CRM and Business Applications with Salesforce Agentforce 360

As the leader in the CRM market, Salesforce has recently developed a comprehensive AI architecture called Agentforce 360 by integrating artificial intelligence end-to-end across its platform. Agentforce 360 can be viewed as an evolved form of Salesforce’s customer data platform, long known as Customer 360. This new platform aims to deliver service from a 360-degree perspective by creating real-time connections among people, AI agents, applications, and data. According to Salesforce, Agentforce is a new way of operating AI across the enterprise by supporting every employee, department, and business process with an “always-on digital workforce.” In other words, Agentforce 360 behaves like an operating system that standardizes human–AI collaboration at enterprise scale; just as an organization has a workforce of employees, it also enables AI agents to participate in workflows.

To realize this vision, Salesforce designed its platform as a four-layer architecture. The first layer is the enterprise data backbone centered on Data 360 (formerly Data Cloud). In 2025, Salesforce significantly strengthened Agentforce 360’s data capabilities by integrating the cloud data management company Informatica. Informatica’s Intelligent Data Management Cloud (IDMC) solution was integrated with Agentforce and Data 360 to provide AI agents with the rich metadata and enterprise data lineage information they need. Analysts evaluate this move as a critical step toward enabling Salesforce’s AI agents to master all enterprise data assets rather than being limited to CRM data alone. Salesforce’s GM responsible for data platforms, Rahul Auradkar, notes that through this integration, they are getting closer to the goal of “enterprise understanding”: combining Salesforce’s comprehensive metadata model with Informatica’s enterprise-scale data catalog to create a “fully fledged data index.” This foundation gives AI agents access to core enterprise business data and the relationships among them, providing contextual awareness. For example, when preparing an offer for a customer, an AI sales agent can simultaneously be aware not only of communication history in CRM but also of inventory status in ERP, delays in the supply chain, or even external market data. The second layer of Agentforce 360 is Salesforce’s two decades’ worth of business logic and workflow library (embedded best practices in sales, service, marketing, e-commerce, etc.). In the third layer are the Agentforce Command Center, where enterprises can create and manage their own AI agents, and the low-code Agentforce Builder tools. The fourth layer is the execution environment where these agents are safely deployed. Thanks to this architecture, Salesforce has managed to keep its platform open and extensible; companies can even bring third-party AI agents from OpenAI, AWS, Azure, GCP, or Oracle into this ecosystem if they wish.

One of the notable features of the Salesforce Agentforce 360 platform is its focus on manageability and predictability. The Agent Script tool, announced in 2025, aims to prevent unexpected outputs by enabling managers to define AI agent behaviors with “if/then” constructs. The Agentforce Builder environment introduced in the same period enables AI agents to be created, tested, trained, and deployed from a single place. Salesforce is also making the popular collaboration tool Slack an integral part of the platform. Within Agentforce 360, Slack is no longer merely a messaging app; it is positioned as an AI-powered work hub: AI modules for areas such as Sales, IT, and HR are integrated into the Slack interface, and Slackbot becomes a personalized assistant that learns user habits and can pull information by searching external sources like Gmail, Outlook, and Dropbox.

Salesforce’s agentic AI move is also a strategic step in its competition with other major market players. Toward the end of 2025, Google announced its Gemini AI Enterprise platform, and Anthropic announced Claude Enterprise, entering the enterprise AI arena. Salesforce stated that with the first release of Agentforce 360, it reached 12,000 enterprise customers and that it is ambitious in its strategy of integrating AI into its products. This figure indicates strong demand for AI agents and suggests that Salesforce’s already broad customer base is rapidly adopting AI capabilities.

Perhaps the most striking example in the Salesforce world is the company’s transformation of its own CRM support operations with agentic AI. Throughout 2025, Salesforce reorganized a 9,000-person customer support unit in favor of AI agents. According to information shared by Marc Benioff in September 2025, AI agents became capable of solving half of customer support requests end-to-end, and as a result, the number of support staff decreased from 9,000 to 5,000 in 12 months. In other words, automation through AI took over a workload equivalent to 4,000 people, reducing costs and increasing response speed. The remaining support teams began to focus only on complex cases, taking on more supervisory and value-adding roles. Salesforce’s internal transformation demonstrates how radically agentic AI can increase productivity when applied correctly. The success has created a domino effect across the industry, as many companies adopt similar AI-supported support strategies. Salesforce’s experience provides a concrete example of rebalancing the workforce with AI. This transformation has also led to the emergence of new job roles. For instance, positions such as “AI Workforce Manager,” “Conversation Quality Lead,” or “Knowledge Curator” began to be defined within companies to monitor AI agent outputs, set human-in-the-loop thresholds in conversations, and keep the knowledge base up to date.

Large organizations are closely following Salesforce’s Agentforce 360 move and SAP’s Joule AI leap in Turkey. While a transformation at this scale has not yet been reported in Turkey, local companies have begun testing SAP and Salesforce’s new AI capabilities in pilot projects. Indeed, in October 2025, the Turkish technology media platform Swipeline covered the Salesforce Agentforce 360 announcement in detail for Turkish readers ahead of the Dreamforce conference. The report highlighted innovations such as Agent Script and Slack integration, noting that Salesforce’s AI strategy aims to build an infrastructure integrated across all products, and emphasizing that the platform’s first release reached 12,000 customers. On the SAP side, SAP Türkiye has been encouraging Turkish customers to adopt these innovations by presenting Joule AI’s capabilities and success examples at events. It is known that large-scale Turkish companies—especially in sectors such as finance, telecom, and manufacturing—are already drawing roadmaps to integrate AI into ERP and CRM processes. This indicates that, as a reflection of the global trend, agentic AI-based solutions are likely to become widespread in Türkiye over the next few years.

Effects on Data Integration and Workload Automation

For agentic AI to succeed, data integration and workload automation play critical roles. AI agents are only as intelligent and effective as the data that feeds them; therefore, siloed, disconnected, or low-quality corporate data is one of the biggest obstacles to agentic AI projects. Indeed, research shows that the failure rate of AI projects is very high, and the reason is often not the algorithm itself but data gaps or inconsistencies. Salesforce’s Informatica integration mentioned above is an example of efforts to address this problem: the CRM giant strengthened metadata and data lineage management so that agentic AI agents can fully access “enterprise memory,” enabling AI to develop a consistent understanding across the entire business rather than being limited to the data of a single department. Contextual awareness is considered the new currency of the agentic AI era—an AI agent must go beyond fragmented or stale data and grasp the context of events in real time to make truly intelligent decisions. As Salesforce’s data executive Auradkar puts it, an agent needs to know “what is happening right now,” and this depends on real-time integration and signal tracking.

Unfortunately, the current state of enterprise systems often falls short of meeting these needs. Traditional data architectures typically rely on ETL (Extract, Transform, Load) processes and data warehouses, meaning data is moved and transformed periodically. Agentic AI, however, requires continuous scanning, indexing, and analysis, so the classic ETL approach may be insufficient. According to Deloitte’s 2025 technology trends research, nearly half of organizations struggle with data discovery and reuse; existing data structures do not directly provide the business context AI agents need for decision-making. To overcome this, pioneers are moving toward making enterprise data discoverable through a Google-like search and knowledge graph infrastructure. SAP’s Business Data Cloud approach and use of the SAP Knowledge Graph are part of this vision: harmonizing data across different applications and presenting it as a single meaningful whole enables Joule agents to make reliable decisions. Agentforce 360’s Informatica-supported data catalog similarly aims to create an AI-ready enterprise golden record by integrating scattered data sources.

While data integration forms a strong foundation, the real value of agentic AI emerges in workload automation. What is meant here is the transfer of processes that human employees routinely perform—time-consuming processes with specific decision points—to AI agents. For example, the Bosch case classification agent can handle the categorization and routing work that customer representatives or managers would normally deal with. Likewise, tools such as SAP’s Offer Analysis Agent can automatically evaluate supplier offers in the supply chain with far more extensive criteria than a human could, and recommend the optimal choice. Such automation takes the workload off employees so they can focus on strategic thinking, relationship management, or creativity. The shift of half of the workload in Salesforce’s support unit to automation enabled the remaining staff to focus on more qualified work, leading to both productivity and employee satisfaction gains.

On the other hand, building management and control mechanisms are also essential in workload automation. Organizations want autonomous systems to remain within certain constraints when making decisions and to consult humans when appropriate. For this reason, both SAP and Salesforce have added strong governance layers to their agentic AI solutions. SAP’s LeanIX AI Agent Hub solution makes it possible to monitor all AI agents used within a company via a central dashboard, track performance KPIs, and intervene when needed. Similarly, Salesforce’s Agentforce command center provides a governance framework that enables the orchestration of different AI agents and delegation from one task to another when necessary. Deloitte analyses emphasize that many organizations initially make the mistake of seeing agentic AI as a simple add-on to existing processes; the real value emerges by designing processes from the ground up to fit an agentic approach. In other words, to make workload automation successful, organizations must move beyond “automating the existing job as it is” and instead ask: “How would AI do this job best, and what should the role distribution look like?” Otherwise, what is referred to as “pseudo-agentification” may occur—classic automation capabilities being presented as if they were agents—and the expected efficiency is not achieved. Poorly designed AI applications can even add workslop to workflows, slowing processes instead of accelerating them. Therefore, alongside data integration and automation, correct process design and a continuous improvement loop are vital to the success of agentic AI projects.

Project Manager Perspective: Managing Agentic AI Projects

Projects that include agentic AI differ from classical software projects in certain respects. Managing these projects involves a dual challenge: both preparing the technical architecture and managing organizational change. For CIOs, CTOs, and project managers, a methodological and strategic approach is essential:

• Preparing the Technology Infrastructure: Agentic AI implementations often require microservices architectures, real-time data flows, and strong API integrations. Gartner warns that “by 2027, more than 40% of agentic AI projects will fail because legacy systems cannot support modern AI requirements.” This brings onto the agenda upgrading existing ERP/CRM infrastructures to work with agentic agents. Project managers should identify legacy systems that may create bottlenecks in AI integration; if possible, they should overcome these obstacles via modernization, or otherwise via temporary solutions or interfaces. Real-time processing capacity, up-to-date APIs, modular architecture, and strong identity/authorization management are the building blocks of agentic integration. Another technical requirement is establishing enterprise knowledge graphs and data catalogs. Project teams should organize enterprise data into knowledge graphs that AI agents can consume, and deploy metadata systems that can track the source and reliability of data. This preparation phase is essential for AI to know “what to look for, and where.”

  
  

• Redesigning Processes: One of the biggest mistakes when implementing agentic AI projects is trying to automate existing business processes as they are without changing them at all. However, AI agents operate differently from humans; steps that make sense for a human may be unnecessary or inefficient for an AI. For this reason, business processes must be reconsidered in a way that fits the agentic environment. For example, a human-centric approval process can be sequential and time-consuming; in contrast, with multiple AI agents working in parallel, this process can be redesigned. Leading organizations are rebuilding process flowcharts from scratch in agentic projects and seeking answers to the question: “How would AI do this process better?” Henry Ford’s famous quote fits well here: “Searching for a better way to do a job that should not be done at all is perfecting something useless.” True transformation comes from eliminating unnecessary work and redesigning necessary work according to human–AI collaboration.

  
  

• Gradual Pilots and Scaling: Agentic AI carries uncertainty alongside high potential. Therefore, projects should begin with small-scale pilots, and scaling plans should be built progressively by measuring the results. Deloitte’s research shows that 30% of companies are in the exploration phase for agentic options, 38% are in the pilot phase, but only 14% are ready to implement, and 11% have solutions actively in production. Moreover, 42% of companies are still in the process of creating an agentic AI roadmap, and 35% have not defined any strategy at all. These data points underscore the importance of acting in a planned manner. Project managers should define a clear roadmap from the start: which processes, which types of AI agents will be deployed, which metrics will measure success, failure criteria, and fallback plans. In addition, KPIs such as productivity, speed, and cost savings obtained in pilots should be presented transparently to top management to ensure continued support for the project. The focus on ROI (return on investment) must remain front and center— as Deloitte notes, defining a concrete ROI expectation approved by finance and business sponsors for each AI initiative is a discipline that prevents AI trials from spiraling into resource waste.

  
  

• Change Management and the Human Factor: Agentic AI projects are not only technology projects; they are also change management projects. As automation increases, ways of working and role definitions within the organization will change. Project managers should address this human dimension proactively. Communication and training are key here. Employees should be told that AI agents are not competitors to their jobs, but support— for example, it should be emphasized that AI will take over routine tasks, enabling them to focus on more creative or strategic work. As seen in the Salesforce example, while some roles decrease quantitatively, new areas of expertise emerge (AI workforce manager, data curator, etc.). Companies should plan training programs to equip employees with new capabilities and instill a culture of working alongside AI. In addition, the ethical and legal dimensions of AI decisions must also be considered—policies and controls (e.g., AI ethical guidelines, model approval processes) should be implemented regarding algorithmic bias, explainability, and compliance. Project managers should work with legal and HR units to ensure AI use aligns with company values and legal regulations.

  
  

• Top Management and Stakeholder Support: Agentic AI projects often cross departmental boundaries and require transformation across the entire organization. Therefore, without ownership from the top, the chances of success decline. McKinsey’s 2025 research shows that in less than 30% of companies, the CEO directly sponsors the AI agenda. Yet in such a strategic transformation, leadership at the CIO, CTO, and even CEO level can make a critical difference. Project managers should secure sponsorship by presenting top management with a clear vision and business value. Similarly, the inclusion of business unit leaders is essential for identifying the right use cases and ensuring adoption. By forming cross-functional teams, IT specialists and operational experts should be brought to the same table, and the AI solution’s practical fit should be designed alongside its technical feasibility.

  
  

In summary, managing agentic AI projects requires the combination of technical readiness, process renewal, incremental scaling, human capital adaptation, and strong leadership. Companies that pay attention to each of these areas can overcome the high early-stage failure rates and reach the productivity and competitive advantage promised by agentic AI. It must be remembered that implementing agentic AI is not a destination but a continuous journey— as models and tools learn and evolve, companies will need to continuously improve their strategies and processes as well. The role of the project manager is precisely to sustain this continuity of change and prepare the organization for the future working model.

Current Examples and Success Stories in Agentic AI Use

The benefits of agentic AI in ERP and CRM integration have moved beyond theory and are showing themselves in practice. Some example cases from around the world reveal the tangible results of this transformation:

• Bosch (Global) – SAP Service Cloud & Joule AI: The Bosch example discussed above shows the gains achieved by integrating SAP’s AI agents into customer service processes. Thanks to the case classification AI within the SAP Joule infrastructure, Bosch minimized human intervention in routing support requests to the right person. As a result, hundreds of manually rule-based routing scenarios were eliminated, errors decreased, and response speed to customer requests increased significantly. Bosch executives confirmed that agentic AI had a directly positive impact on customer experience by describing the solution as a “game changer.”

• Salesforce (Global, Internal Use) – Agentforce 360 & Autonomous Agents in Customer Support: Salesforce’s deployment of Agentforce 360 agents in its own customer support organization achieved a 50% automation rate within a year, successfully transferring an equivalent workload of 4,000 people to AI. This transformation reduced operational costs significantly while maintaining customer satisfaction. In addition, by creating new roles for AI oversight and optimization, employee job security and productivity were managed together. This case has served as an example for many businesses because it underlines that agentic AI can become a scalable success story.

• PwC Germany (Global) – Real-Time AI in Financial Reporting: PwC Germany established real-time, automated reporting systems in its finance department using SAP’s Business Suite and Business AI solutions. CFO Stefan Frühauf notes that financial close and reporting processes, which previously relied on static reports, became interactive and continuously updated thanks to AI-supported analytics. This enables executives to monitor instant financial indicators and intervene when necessary, rather than waiting for month-end reports. This example shows that agentic AI can also enable the shift from static to dynamic in financial decision-support systems.

  
  

The examples above show the tangible benefits agentic AI provides across different sectors and use cases. Global giants report operational improvements, cost savings, and enhancements in customer experience thanks to dynamic decision support and automation. These success stories also contain valuable lessons for other businesses: With well-chosen use cases, top management support, a strong data foundation, and well-managed change, agentic AI investments can bear fruit in a short time.

The Future Operating System: Humans + Agent Workforce

Agentic AI has begun to change the rules of the game in areas such as ERP and CRM, which can be considered the corporate nervous system. Static, retrospective perspectives are giving way to proactive management based on real-time data. For executives at the CIO, CTO, and CEO level, this development is not merely a technology trend, but a business strategy matter. A properly implemented agentic AI integration can increase decision-making speed, improve customer satisfaction, and deliver dramatic gains in operational efficiency. However, this transformation requires visionary leadership and organizational alignment as much as technology investment.

This wave of transformation—shaped globally by leading platforms such as SAP and Salesforce—also affects organizations in Turkey. Turkey offers suitable ground for adopting agentic AI practices thanks to its young and dynamic workforce, technology adoption speed, and large-scale organizations. Over the next few years, we will witness a rapid increase in agentic AI-based project examples in both the public and private sectors in Turkey. In sectors such as banking, telecom, manufacturing, and retail—where data and customer transactions are dense—the inclusion of autonomous AI agents into business processes seems inevitable. Of course, organizations must take cautious but determined steps on this journey. Data security and privacy, transparency and ethical compliance in AI decisions, and regulatory compliance should remain high on management teams’ agendas. With the “first, not harm” principle, it is essential to explain the change brought by agentic AI correctly to employees and customers and to run transparent change management.

As a result, leaders who include agentic AI integration in their strategic plans today will be one step ahead in tomorrow’s competitive world. Organizations that embrace the shift from static reports to dynamic decisions, use data as a living asset in real time, and effectively manage human–AI collaboration will move ahead of their competitors in digital transformation. At every step of this comprehensive transformation journey, the methodological approaches and success stories shared in this article will guide. When managed correctly, agentic AI is poised to become one of the strongest allies enabling enterprises to navigate future uncertainties in both the global arena and the Türkiye market.

References

• Verma, Anushree. “Why Agentic AI Projects Fail—and How to Set Yours Up for Success.” Harvard Business Review, 21 October 2025.

• Plumb, Taryn. “Salesforce’s Agentforce 360 gets an enterprise data backbone with Informatica’s metadata and lineage engine.” CIO, 9 December 2025.

• Rowan, Jim. “The agentic reality check: Preparing for a silicon-based workforce.” Deloitte Insights – Tech Trends 2026, 10 December 2025.

• Yeung, Ken. “SAP Recasts Joule as an Agentic AI Platform Driving Workflow Automation.” The Letter Two (blog), 6 October 2025.

• Yaka, Ezgi. “Salesforce’un yeni kurumsal yapay zeka platformu: Agentforce 360.” Swipeline, 13 October 2025.

• SAP. “Joule Copilot from SAP | Artificial Intelligence.” SAP official product page. Accessed 14 December 2025.

• Dmitry V. “The AI Workforce Transformation: Lessons from Salesforce’s Agentic AI Implementation.” CTimes.tech, 22 October 2025.

• SAP Customer Story – Bosch: Florian Haustein’s evaluation regarding SAP Joule AI agents (Bosch success story, SAP Press, 2025).