Businesses
AI Agents for Business: How CEOs Are Automating Workflows in 2026
09 Nov 2026
From Siemens' automated sales qualification to ServiceNow's AI-powered customer support, businesses are moving beyond chatbots to systems that execute tasks, coordinate workflows, and deliver measurable operational results.
Artificial intelligence is beginning to change how companies organise and execute everyday work. Businesses are deploying AI agents that respond to customers, qualify sales enquiries, process support tickets, and complete tasks that previously required employees to move between several software systems.
The shift is already visible in major enterprises. According to McKinsey's State of AI in 2026, published in August, 40% of respondents from organisations generating more than $1 billion in annual revenue reported scaling AI agents in at least one business function, up from 27% the previous year. Among smaller organisations, the proportion remained at 22%.
However, wider adoption has not automatically translated into stronger financial performance. Only 37% of respondents reported that AI had contributed to their organisation's earnings before interest and taxes (EBIT), while approximately 6% qualified as AI high performers under McKinsey's definition.
For CEOs, these findings raise an important question: where can AI agents deliver measurable business value, and what does it take to move from isolated automation to reliable operational improvement?
What Are AI Agents, and How Do They Differ from Traditional Automation?
Traditional automation generally follows predefined instructions. A conventional workflow might automatically send an invoice when a purchase is completed or assign a support ticket according to a fixed set of rules.
AI agents can go further. Depending on their design and permissions, they can interpret a request, retrieve information, decide which permitted tools to use, and complete a sequence of actions to achieve a defined objective. Enterprise software providers are expanding these capabilities, as seen in the Oracle and OpenAI partnership, which aims to integrate AI capabilities into business applications and workflows.
A customer service agent, for example, might identify a customer's problem, check an order record, retrieve a company policy, and initiate an authorised resolution. More complex situations can be passed to a human employee.
The difference matters because many business processes involve changing information, exceptions, and decisions that cannot be handled through simple rules alone.
Yet autonomy is not unlimited. AI agents remain dependent on the quality of their data, available integrations, operating instructions, and the controls placed around their actions. An agent that answers questions accurately may still be unsuitable for independently changing financial records or approving sensitive transactions.
The business case therefore depends on matching the agent's capabilities to the operational task and the level of risk involved.
Siemens: Automating Sales Qualification Across 132 Countries
German industrial technology company Siemens provides one example of how AI agents are being introduced into complex B2B sales processes.
In September 2026, Salesforce published details of Siemens' deployment of two Agentforce-powered agents designed to engage and qualify inbound sales leads. Previously, identifying promising opportunities required sales professionals to research prospects and assess their purchasing requirements manually.
The first agent contacts incoming leads and manages initial engagement. A second qualification agent gathers information about budget, purchasing authority, needs, and purchasing timelines before routing suitable opportunities to sales representatives.
According to Salesforce's published customer case study, the system provides an initial response to approximately 2,500 in-scope leads each month across 132 countries. Lead response times have fallen from days to minutes, with an 11% engagement rate and a 6% qualification rate reported for the deployment.
The distinction is that agents manage the initial qualification process while human sellers retain responsibility for more complex commercial discussions.
For CEOs managing large sales teams, this example demonstrates how automation can reduce administrative delays while allowing employees to concentrate on prospects who require specialist attention.
Customer Service: AI Agents Move from Answers to Resolution
Customer support has become another area where organisations are demonstrating measurable outcomes.
In October 2026, managed services provider Thrive and ServiceNow announced results from an AI-powered service transformation. Thrive deployed specialised agents and workflow automation to manage incident routing, ticket investigation, and support processes.
ServiceNow reported that the company had achieved a 73% first-contact resolution rate, alongside a 63% reduction in time to resolution. The platform also routed more than 315,000 tasks over six months, with an estimated 21,000 hours of manual work saved.
These results reflect a broader deployment of ServiceNow technology, AI agents, and automation rather than the isolated effect of a single AI model.
The example illustrates how AI can support technical service teams by reducing repetitive ticket processing and improving access to information needed to resolve customer problems.
Canada's Bell has also deployed ServiceNow AI agents in customer case and complaint management. The agents check incoming cases for completeness, populate relevant fields, and identify possible duplicates.
ServiceNow reported in March 2026 that Bell had achieved a 25% improvement in customer response time following the deployment.
Both examples show the operational value of connecting AI agents to existing workflows. Rather than generating answers alone, the systems help employees move cases towards resolution.
Sales, Marketing, and Customer Engagement
AI agents are also changing how businesses respond to potential customers before a formal sales conversation begins.
Salesforce has implemented Agentforce on its own websites to help visitors find information about products and pricing. The system answers questions and connects relevant enquiries to sales representatives.
In a customer story published in May 2025, Salesforce reported more than 100,000 conversations initiated since launch, more than 30,000 conversations influencing lead generation, and a 40% year-on-year reduction in qualification time.
These figures are company-reported results. They show the scale of the deployment but should not be interpreted as evidence that all organisations will achieve comparable conversion improvements.
For marketing and sales leaders, the opportunity lies in connecting website engagement, customer information, and qualification processes. An AI agent may help respond to enquiries outside normal business hours, gather relevant context, and pass structured information to the appropriate team. This shift also extends to digital commerce, where AI agents are changing online shopping, business sales, security, and payments, creating new ways for businesses to interact with customers throughout the purchasing process.
However, the quality of these interactions depends on accurate product information, properly maintained customer data, and clear limits on what an agent is authorised to communicate or promise.
Where Else Are Businesses Using AI Agents?
Beyond sales and customer support, organisations are exploring agent-based workflows across several business functions.
In IT operations, agents can classify incidents, retrieve troubleshooting information, and prepare cases for technical teams. Human approval remains appropriate before high-impact actions, such as changing production systems or access permissions.
In human resources, agents can assist with onboarding requests, answer questions using approved policies, and coordinate routine administrative tasks between HR and IT systems.
Finance departments can use suitably controlled agents to support document processing, reconciliation preparation, and exception identification. Sensitive approvals, payments, and changes to financial records require stronger safeguards.
Software engineering is another area of adoption. McKinsey's 2026 survey found that approximately one in five respondents reported scaling software coding agents, with the share reaching 31% among larger enterprises.
These applications do not all represent the same level of autonomy. Some systems prepare recommendations for employees, while others can perform narrowly defined actions without requiring approval at every step.
CEOs need to understand that distinction before evaluating the business benefits.
Why AI Agents Still Struggle to Deliver Enterprise-Wide ROI
The growing number of deployments does not mean that AI transformation has become straightforward.
McKinsey's August 2026 findings indicate that employee-level productivity improvements are more widely reported than organisation-wide financial gains. Eighty percent of respondents said AI had improved their individual productivity, while only 37% attributed some EBIT impact to AI use.
Cost is also becoming a constraint. Approximately one in five respondents reported that AI-related operating expenses, including model usage costs, had restricted their adoption.
A business might automate hundreds of tasks without reducing overall costs if it simultaneously increases software expenditure, creates additional review work, or fails to redesign the surrounding process. As explored in why some companies succeed with agentic AI while others remain stuck in pilot projects, access to reliable, real-time data and the ability to integrate AI into existing operations can influence whether deployments progress beyond experimentation.
Successful adoption therefore requires more than purchasing an AI platform. Organisations must identify where employees lose time, which operational outcomes matter, and how automated work will be measured.
Microsoft's 2026 Work Trend Index reinforces the importance of organisational readiness. Based on a survey of 20,000 AI-using workers across ten countries and analysis of Microsoft 365 usage patterns, the report highlights the need for clearer work processes, quality standards, and human oversight as agent adoption expands.
For business leaders, the implication is that AI adoption must be supported by changes in how work is designed and managed.
AI Agent Governance: Why Identity, Permissions, and Oversight Matter
As AI agents gain access to business applications and customer information, they also introduce new operational and cybersecurity risks.
In May 2026, the US Cybersecurity and Infrastructure Security Agency (CISA), working with international cybersecurity partners, published guidance on the careful adoption of agentic AI services.
The guidance identified risks including excessive permissions, unintended system behaviour, and gaps in accountability. It recommended limiting agent autonomy, strengthening identity management, maintaining oversight, and conducting regular security assessments.
An AI agent that reads customer information presents a different risk from one authorised to modify accounts, issue refunds, or communicate with external systems. New approaches to enterprise AI agent security, including NVIDIA's Open Agent Safety Platform, aim to strengthen the controls governing autonomous systems and their access to business resources.
Before granting those capabilities, companies should establish who owns the agent, which systems it can access, what actions it can perform, and when human approval is required.
Organisations should also maintain records of agent actions, monitor failures, and ensure that access can be revoked when necessary.
These measures make it possible to introduce automation while preserving accountability for decisions affecting customers, employees, and business operations.
A Practical Framework for CEOs Introducing AI Agents
Businesses considering their first production deployment can begin with a limited, measurable workflow rather than attempting to automate an entire department.
The first step is to identify a recurring process with a clear operational problem. Long customer response times, delayed sales qualification, or repetitive support administration can provide suitable starting points.
Next, establish a baseline. A CEO should know how long the current process takes, what it costs, how often errors occur, and whether customers or employees are satisfied with the outcome.
The organisation can then introduce an agent with restricted permissions and defined escalation rules. Early testing should examine not only whether the agent completes tasks but also whether the results are accurate and appropriate.
Performance should be measured through operational indicators such as resolution time, cost per completed task, qualified leads, error rates, customer satisfaction, and the number of cases requiring human intervention.
A successful pilot can provide evidence for expansion. A failed pilot can reveal weaknesses in company data, process design, or system integration before those weaknesses affect a larger operation.
The decision to scale should follow demonstrated performance, not the number of AI agents created.
The CEO's Role in an Agentic Business
AI agents are becoming part of how companies organise sales, service, IT, and internal operations. The experiences of Siemens, Thrive, Bell, and Salesforce show that specific workflows can benefit from faster responses, reduced manual processing, and more consistent information handling.
The next stage of adoption will depend on whether businesses can connect these operational improvements to broader commercial results.
CEOs remain responsible for choosing which activities should be automated, determining how employees and agents work together, and ensuring that customers can trust the outcomes.
The most useful starting point is a measurable business problem, a clearly defined workflow, and an accountable team responsible for its performance.
As agentic systems become more capable, the question for business leaders is no longer whether AI can complete individual tasks. It is which tasks should be delegated, under what conditions, and how their contribution to business performance can be demonstrated.
Sources:
- https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai
- https://www.microsoft.com/en-us/worklab/work-trend-index/agents-human-agency-and-the-opportunity-for-every-organization
- https://www.salesforce.com/customer-stories/siemens/agentic-sales-leads/
- https://newsroom.servicenow.com/press-releases/details/2026/Thrive-scales-customer-experience-with-ServiceNow-AI-achieving-73-first-contact-resolution/default.aspx
- https://newsroom.servicenow.com/press-releases/details/2026/Mobile-World-Congress-2026-Autonomous-CRM-puts-AI-to-work-in-telecom/default.aspx
- https://www.salesforce.com/customer-stories/agentforce-for-dot-com/
- https://content.govdelivery.com/accounts/USDHSCISA/bulletins/41544ff
Share
Sara Srifi
Sara is a Software Engineering and Business student with a passion for astronomy, cultural studies, and human-centered storytelling. She explores the quiet intersections between science, identity, and imagination, reflecting on how space, art, and society shape the way we understand ourselves and the world around us. Her writing draws on curiosity and lived experience to bridge disciplines and spark dialogue across cultures.
