How IT leaders use agentic AI for business workflows

“We’re not just automating a handful of manual tasks and processes across a department or two,” says Kellie Romack, CDIO at ServiceNow. “We’re infusing AI agents everywhere to reimagine how we work and drive measurable value.”

Agentic AI is the new frontier in AI evolution, taking center stage in today’s enterprise discussion. AI agents topped Forrester’s 2024 trend list, and Salesforce expects one billion in use by the end of fiscal year 2026.

Though loosely applied, agentic AI generally refers to granting AI agents more autonomy to optimize tasks and chain together increasingly complex actions.

UIPath’s 2025 Agentic AI Report surveyed US IT execs from companies with $1 billion or more in revenue and found that 93% are highly interested in agentic AI for their business. The study found better oversight of business workflows to be the top perceived benefit of it.

“Using AI effectively is now a fundamental expectation of everyone at Shopify,” added Shopify CEO Tobias Lütke in a leaked internal memo. Many organizations are in the process of moving AI hype into calculated action. So what are these specific workflows that more autonomous AI can supercharge?

Use cases for AI agents span countless business workflows. “AI agents are valuable across sales, service, marketing, IT, HR, and really all business teams,” says Andy White, SVP of business technology at Salesforce. And executives see a high potential in streamlining the sales funnel, real-time data analysis, personalized customer experience, employee onboarding, incident resolution, fraud detection, financial compliance, and supply chain optimization.

Enriching the sales pipeline

Jay Upchurch, CIO at SAS, backs agentic AI to enhance sales, marketing, IT, and HR motions. “Agentic AI can make sales more effective by handling lead scoring, assisting with customer segmentation, and optimizing targeted outreach,” he says.

And Eilon Reshef, co-founder and chief product officer for revenue intelligence platform Gong, says AI agents are best deployed as a well-defined task interwoven into a larger workflow. An AI briefer could inform a sales pipeline review process, for instance, or an AI trainer could simulate customer interactions as part of an onboarding program, he adds.

Another area is democratizing data analysis and reporting. Cloud-based data-warehousing company Snowflake has taken this to heart with their Sales Assistant, an internal agentic AI tool designed to empower their global sales team with instant, data-driven insights, resulting in time savings and improved targeting.

“By enabling real-time, autonomous data retrieval, analysis, and visualization through natural language queries, users can accelerate time-to-insights and reduce dependency on centralized data teams,” says chief data and analytics officer Anahita Tafvizi.

So agentic AI use cases can span the entire sales pipeline — well-suited for qualifying leads, triggering personalized offers, and more. “I’m really keen to see how agentic AI is suited for driving sales conversions by enabling sales teams to strategically target clients offering the highest potential returns,” adds Rebecca Fox, group CIO at NCC Group, a large cybersecurity consulting firm. She sees potential in using agents to schedule client work and match client requirements with the best-skilled and cost-effective resources.

Customer gains

Customer experiences are well-suited for an agentic boost as well. “It’s no secret AI agents are revolutionizing the way we work and the way we interface with customers,” says SAS’s Upchurch.

One area is personalizing on-page digital interactions. “Agentic AI can quickly guide consumers to the right product for their needs, or based on consumer reviews,” says Velia Carboni, CIO of SharkNinja, a global product design and technology company. “By leveraging AI-driven solutions, we aim to provide an engaging and intuitive shopping experience.” Salesforce’s Agentforce plays a key role in SharkNinja’s digital transformation, too, with agentic AI being evaluated across user browsing, product selection, recipe discovery, and customer support.

For some users, the net gains from AI agents could be life-altering. “We’re harnessing the power of agentic AI to enhance our advising workflows,” says Siva Kumari, CEO of College Possible, a nonprofit that supports students on their path to college that uses AI agents to analyze students’ needs, surface relevant institutions, and synthesize information quickly and accurately. By reducing research time from 35 minutes to under three, coaches can now build better relationships and improve their one-on-one guidance. The result is speed, personalization and, hopefully, more successful college admissions.

Streamlining customer support

AI agents are useful for automating repetitive support inquiries, says Salesforce’s White. He points to how 1-800Accountant, a large US virtual accounting firm, uses AI agents to deflect 65% of incoming status requests. “By automating these tasks, their human coworkers can focus on more complex client needs,” he says.

Agentic AI is a natural extension for call centers as well since they already often deploy AI for areas like voice-to-text transcription, real-time multilingual translations, and sentiment analysis, says Luiz Domingos, CTO and head of large enterprise R&D at Mitel, a provider of call and contact center software.

“We’re implementing AI-powered agentic tools like autonomous virtual agents with chat and speech to handle and resolve customer inquiries and gather information in case an escalation to a live agent is required,” he says. Agentic AI is also being used for real-time support, CRM integration, and multilingual communication, he adds.

Automating backend tasks

Every corner of the business is touched by agentic AI, and many companies are in the process of implementing it to improve various backend business operations. One specific example is order processing. “We really like agentic AI to handle routine order processing scenarios,” says Brian Glass, CIO at Transcend Company, a platform for health service providers. AI agents assist there by validating requirements and compliances when ordering controlled substances and medications. “We’re looking at areas where we can remove routine, repeatable work and give our employees time back in their day,” Glass adds.

ServiceNow’s Romack similarly sees AI agents as ways to unlock human potential across business domains. At ServiceNow, they’re infusing agentic AI into three core areas: answering customer or employee requests for things like technical support and payroll info; reducing workloads for teams in IT, HR, and customer service; and boosting developer productivity by speeding up coding and testing.

Automation is essential in lean nonprofit work and Good360, the sixth largest charity in the US, has deployed assistive AI agents to improve how they match donated goods with nonprofit partners. “One agent supports daily operations while another helps our disaster recovery team quickly align products with crisis-response organizations,” says CTO Stephane Moulec. “High-volume, repetitive tasks are ideal for AI.”

Simplifying financial workflows

Industries that manage high-volume, data-intensive workflows such as financial services, telecommunications, and healthcare are primed to benefit from agentic AI, says David Vidoni, CIO at Pegasystems, an AI decisioning and workflow automation platform. “These sectors rely on complex, multi-step processes to serve customers efficiently while ensuring compliance.”

In general, Vidoni foresees agentic AI playing a key role in modernizing legacy workflows with increased levels of automation. “Across industries, any process that involves repetitive decision-making, data validation, or cross-system orchestration can benefit from automation, leading to greater efficiency and cost reduction,” he adds.

Similarly, software provider Akamai is prioritizing agentic AI where processes are already highly matured and supported by high-quality data and security controls. “Any workflow that’s rules-based, data-heavy, or requires fast decision-making is a prime target,” says Kate Prouty, the company’s SVP and CIO, who adds that finance is of specific interest.

Financial operations are full of repetitive, rules-based tasks, which agentic AI could streamline.

“Right now, we’re focused on rolling out agentic AI for contract automation,” says Milind Shah, CTO of Xerox IT Solutions. “Think summarizing, reviewing, even flagging risk across thousands of documents. We’re also exploring agents to handle internal IT support tickets and finance workflows like invoice matching and spend analysis.”

As Xerox continues its reinvention, shifting from its traditional print roots to a services-led model, agentic AI fits well into that journey. In addition to contract management, Shah highlights procurement and IT support as areas where AI agents can cut through the noise and act autonomously.

Boosting IT and security

AI agents are transforming software engineering, aiding in code generation, testing, refactoring, observability, and beyond. And other technical areas, like low-code data integration, are set to get a boost as well, and Gartner’s 2024 Magic Quadrant report says that incorporating AI assistants and AI-enhanced workflows into data integration tools will reduce manual intervention by 60%.

“NCC Group is already prioritizing AI in cybersecurity response automation, IT process management, and customer support,” adds NCC’s Fox. “With agentic AI, we see the outcome from this activity just becoming more effective. These areas were chosen for their clear ROI potential.”

Evaluating agentic AI’s impact

Executives tend to agree that agentic AI frees up humans to focus on higher-impact activities like strategic partnerships and relationship-building.

“It’s all about making operations smarter, faster, and more proactive as we scale our services business,” says Xerox’s Shah. The goal, adds NCC’s Fox, is reducing human bottlenecks, increasing accuracy, and enhancing responsiveness to rapidly changing business and customer needs.

Armed with contextual data, agentic AI can surface insights, trends, or anomalies more protectively, helping direct business decisions. “Organizations can move from reactive reporting to dynamic, AI-driven decision-making, driving greater agility and impact,” says Snowflake’s Tafvizi.

Although excitement around agentic AI is palpable, several obstacles remain. A 2025 study from Crane Venture Partners, which surveyed executives representing $3 to $4 billion in annual tech spending, found that over half of respondents cite data integration and interoperability as major hurdles, underscoring the gap between AI ambition and execution. And around 45% also cite data governance and compliance concerns.

So most executives prevent adding further complexity for the sake of following trends. “We’re taking a purposeful and deliberate approach to implementing AI across our organization, and this thoughtful strategy extends to agentic AI workflows,” explains Akamai’s Prouty.

Currently, it’s easy to overpromise while the technology is still developing. “There’s a lot of hype about AI replacing humans but the reality is AI systems aren’t yet capable of carrying out complex multi-step tasks,” says Gong’s Reshef. “Most organizations look for a consistent, repeatable process so they’re not looking for autonomous agents to make independent decisions and improvise processes.”

Since LLMs are non-deterministic, meaning they can provide different responses to the same input, maintaining consistency requires new validation procedures. “Testing is something we’ve been spending a lot of time on,” says Salesforce’s White. “New automation and environments that better match production have been necessary for successful testing.”

For others, integration remains the biggest obstacle. “The biggest challenge with AI agents is getting them to effectively communicate and work together across the tech stack,” says ServiceNow’s Romack. “On the integration side, the real value comes when AI agents can talk to each other, share data, and perform tasks across different systems.”

Many reiterate that the goal is not to replace humans. “Across every use case, our approach remains practical and human-centered,” adds Good360’s Moulec. “We use AI to scale our mission, not replace our people.”

Agentic AI’s future in the enterprise

“In the enterprise, AI agents will go from assistants to decision-makers — predicting problems, taking action, and continuously optimizing operations,” says Akamai’s Prouty. “Expect tighter AI-human collaboration, where AI handles execution, and humans focus on strategy.”

While the future looks agentic, we’re still in the early stages, as Salesforce shows that only 11% of CIOs have fully adopted AI due to technical and organizational challenges, like building AI agents from scratch, even though 84% say AI will be as significant to businesses as the internet.

To help close the gap, tech companies have introduced AI agent capabilities, from Salesforce’s Agentforce to Microsoft’s Magnetic-One, IBM’s Bee Agent Framework, and Google’s ADK framework for multiagent systems. Salesforce also recently introduced an Agentic Maturity Model, offering a clearer path toward AI agents that can act autonomously and collaborate with humans.

Always on, always improving

As the enterprise AI journey unfolds, most IT leaders are taking a measured, iterative approach. Yet the long-term vision is already clear: AI agents won’t just support workflows — they’ll reshape how work gets done.

“I envision a future where thousands of AI agents work in the background, each focused on a narrow task but able to hand off and collaborate seamlessly,” says Romack. “It’ll feel like a team of digital coworkers, always available and always improving.”