As Companies Bet Big on Agentic AI, Strategies Must Shift to Drive Results

As AI capabilities rapidly evolve, enterprises are shifting their priorities from Gen AI toward more advanced applications, while mentions of agentic AI quickly moving to the center of the C-suite agenda.

Reflecting this surge in interest, there has been a sharp uptick in mentions of the word “agentic” on earnings calls and in investor meetings, as tracked by Bloomberg. From 2024 to 2025, mentions of ‘agentic’ increased more than 3,000%, with mentions of ‘generative’ declined over the same period.

As companies advance agentic plans, they can learn from earlier AI applications, says Shibani Ahuja, Senior Vice President, Enterprise IT Strategy at Salesforce. 

Ahuja, who has held leadership roles at major technology firms and writes about AI, digital transformation and enterprise IT strategy, says 2026 could usher in a more mature phase of AI capable of driving measurable ROI—but only if leaders configure their technical architecture to integrate agentic AI at scale.

Building the foundations for agentic AI

Reflecting companies’ AI enterprise deployments so far, a recent Deloitte survey found that many organizations may not have achieved the pace of progress initially envisioned.

For AI to deliver real value, leaders need to move beyond siloed models and shift their focus—and funds—to enterprise-wide integration that allows for better design, clearer economic accountability and the capacity to scale, Ahuja says. 

“The model is the brain, but it requires the rest of the body,” she says. “It requires the rest of the system to be able to actually execute.”

Moving beyond previous AI models requires unprecedented integration across divisions and data. To achieve this broad implementation, Salesforce and AWS have partnered to give enterprise leaders a secure solution that addresses the critical roadblocks to scaling AI, such as trust, governance and time to value.

With the release of Agentforce 360 for AWS, an offering running entirely on AWS infrastructure and powered by Amazon Bedrock models, enterprises have a compliant foundation for AI procurement and agent deployment. The platform addresses common architectural weaknesses that could render adoption insecure, incompatible or inoperable at scale.

The Salesforce and AWS partnership lays the foundation for scaling agentic AI through four core pillars: unified data; agentic security and interoperability; modernized contact centers; and streamlined procurement through AWS Marketplace. Capabilities include allowing agents to access the right data in real time while preserving security, orchestrating actions across multiple applications and providing visibility into exactly what agents are doing. 

Agentforce leverages the key strengths of both the businesses behind it. Salesforce brings a deep understanding of business workflows, customer context and how work actually gets done. As the world’s most broadly adopted cloud platform, with reliability and security already trusted by enterprises, AWS provides the extensive global infrastructure needed to maximize value and translate AI investments into measurable ROI. 

Companies are seeing the results in their workflows. RBC Wealth Management, an investment and private banking firm that’s a division of the Royal Bank of Canada, utilizes Agentforce and AWS to help financial advisors prepare for client meetings. The process, which previously took an hour, now takes under one minute, and results have proven consistent and reliable, with a near-zero hallucination rate. 

Other industries, such as manufacturing and automotive, are automating customer service processes to help improve response times, elevate service and streamline the customer experience. Toyota Motor North America, for example, is planning to implement workflows that will improve appointment scheduling and loaner vehicle management. 

Solving for the “swivel effect”

The customer insights that led to the creation of Agentforce 360 are a window into some of the common concerns that companies have about onboarding agents—including a prevailing concern that their virtual workspaces will sprawl, and fragment attention between multiple task-oriented autonomous entities. 

Ahuja thinks of this concern as a desire to avoid the “swivel effect”—a need to shift between multiple interfaces to manage agents assigned different tasks.

“Users don’t want to have to track multiple agents on multiple screens for separate tasks,” Ahuja says. “They want agents that are going to work across enterprise systems.”

To solve this, Agentforce 360 prioritizes true interoperability, ensuring that agents aren't just isolated task-performers, but are capable of exchanging data and logic across the entire enterprise ecosystem.

Agentforce built on what worked in Salesforce’s original model: streamlining the patchwork nature of customer relationship management into a single platform. Applying the same approach to training, deploying and monitoring AI agents prioritizes enterprise systems over isolated tools, Ahuja says. 

Choosing a side in the build-or-buy debate

Companies that still choose to build agents in-house—usually in an attempt to retain ultimate control—must reckon with what, exactly, they have the capacity to build. Entire models? Entire data centers to “own” the infrastructure? The UI for agent testing and monitoring? The data lakes that agents must access to make decisions and act? And even if enterprises undertake the initial build, many don’t consider the costs of ongoing maintenance and compliance.

“Are you a bank, or are you an AI company?” Ahuja asks. “If the answer is a bank, then you should be focused on digital banking experiences, rather than building out AI infrastructure.”

What Agentforce works to deliver—and what all agentic systems strive for—is seamless agility in agent orchestration, where time to value isn’t delayed by a bespoke build or disrupted later by outdated architecture. 

Architectural integrity is key to realizing returns

Encouragingly, a critical mass of IT leaders now view architecture as underpinning agentic effectiveness. A Salesforce report released in February reveals that 96% of 1,050 IT leaders surveyed agree that agent success depends on seamless data integration across systems.

To live up to the hype, enterprises should focus on system-wide integration and quality architecture, Ahuja says. This will lead to higher-quality agent performance and superior customer interactions that actually move the needle. 

At this transformation inflection point, that’s key to achieving leaders’ agentic ambitions—and meaningful ROI.

“For the first time, leaders are not simply deciding where technology sits,” Ahuja says. “They are deciding how work itself will be redesigned, how decisions will be made and what kind of enterprise they want to become.”