Artificial Intelligence is no longer just a lab experiment within companies. Salesforce’s latest annual CIO survey confirms a turning point: in just one year, the full deployment of AI projects has grown by 282%, moving from 11% of organizations in 2024 to 42% in 2025. The “pilot phase” is over; the era of scaling begins.
But this leap isn’t happening alone. As projects spread throughout organizations, a new bottleneck emerges: trust. Not so much in the models themselves, but in the data fueling them, in security, governance, and the company’s actual capacity to deploy AI without losing control.
The study, conducted with research firm NewtonX based on surveys of 200 CIOs across 24 countries, paints a picture where tech leaders are shifting from mere technical enablers to organizational change architects, internal negotiators, and increasingly, guardians of digital trust.
From Eternal Pilot to the “Era of Scale”
Just a year ago, most CIOs were focused on cleaning data, closing inherited security gaps, and running controlled pilots of generative AI. Today, the landscape has changed drastically:
- Complete full AI deployment jumped from 11% to 42% of surveyed organizations, a 282% increase.
- AI budgets have almost doubled.
- About 30% of AI budgets are now specifically allocated to agentic AI projects—systems capable of executing tasks autonomously to meet business objectives.
- A 96% of CIOs confirm that their company already uses or plans to use agentic AI within the next two years.
The conversation has shifted from “let’s see what ChatGPT can do” to “how do we embed AI agents into critical processes without compromising data, security, or compliance?”
The CIO Goes from Technologist to Change Leader
This expansion of AI has transformed the CIO’s role. The study shows that 75% feel more confident in their position than a year ago, and 61% believe they are ahead of competitors in AI adoption. But this confidence is not solely based on technical knowledge—it’s supported by skills that, just a few years ago, weren’t even associated with the IT function.
According to the survey, CIOs have spent the past year strengthening three key “soft skills” for the agentic era:
- Leadership: 61% have explicitly improved their leadership capabilities.
- Storytelling and narrative: 57% have worked on explaining AI, its limits, and its value to the business, executive committees, and staff.
- Change management and communication: 55% focus on these skills to ensure adoption moves beyond isolated pilots.
Thus, CIOs are moving beyond being “the ones who know servers and clouds” to become translators between technology and business. Their role now involves aligning the CEO and top management, prioritizing use cases, measuring results, and mitigating the cultural shock of introducing AI agents into daily operations for thousands of employees.
Customer Service: The Primary Testing Ground for Agentic AI
If there’s a sector where AI is demonstrating tangible impact, it’s customer service. 65% of CIOs report working more closely with customer support teams than with any other department, placing service at the top in enthusiasm, readiness, and adoption of agentic AI.
Usage data from Agentforce—the Salesforce agents platform—supports this: the average number of customer interactions led by an AI agent has multiplied by 22 in only the first half of 2025.
Why is customer service the ideal testing ground?
- Use cases are clear: ticket classification, FAQ responses, resolution proposals, history summaries.
- Return on investment can be measured in time saved, response time reductions, and customer satisfaction.
- Agents can start as “copilot” assistants suggesting answers before moving to fully automating complete workflows.
What’s happening today in contact centers largely previews what we’ll see elsewhere in the company: agents who not only respond but consult internal systems, update records, and trigger actions in the background.
Scaling AI Without Organizational Alignment: The Major Risk
The report also highlights a concerning gap between what CIOs know they should do and what’s actually happening:
- 81% believe that agentic AI increases the need to work more closely with HR, Finance, or Sales departments.
- However, less than half are regularly collaborating with these departments on AI implementation.
In other words, technology is being deployed, but organizations are not necessarily moving at the same pace. Many projects originate from IT or a specific department (usually service, marketing, or customer experience), without a cross-functional strategy that considers impacts on staffing policies, sales commissions, regulatory compliance, or change management.
Even CIOs acknowledge this: 93% believe that AI agent success depends on integrating it into actual workflows, not just using tools in isolated instances. The step of moving from a platform to redesigning processes around it remains pending in many companies.
Furthermore, 61% of tech leaders prefer investing in familiar vendors integrated into their current ecosystem. While this “security” facilitates faster adoption, it can also lock organizations into inflexible architectures if open standards, interoperability, and future platform changes aren’t also considered.
Data Trust: The New Bottleneck
The report’s headline is clear: trust is the new bottleneck. For CIOs, the two main concerns regarding AI in their companies are:
- Data security and privacy.
- Quality and reliability of data feeding the models.
However, investments often don’t align with these priorities. On average, only 14% of IT budgets are allocated to data security, and just 23% of CIOs feel fully confident that they are investing in AI with robust data governance built-in.
The contrast is stark: companies rush to deploy AI agents in front and back office but leave cracks in data cataloging, quality, traceability, and access control unaddressed. Without this foundation, agentic AI can turn from an asset into an operational risk—decisions based on incomplete data, reproduced biases, and in worst cases, security incidents or leaks that erode customer and regulator trust.
What This Means for CIOs Looking Toward 2026
Salesforce and NewtonX’s study confirms that CIOs have crossed a point of no return: AI is no longer a peripheral option but a core element of their technology and business strategy.
In the coming months, their top priorities will include:
- Accelerating data governance: improving data quality, unifying sources, clarifying responsibilities, and strengthening security.
- Establishing transversal alignment mechanisms: AI committees involving business, HR, legal, finance, and operations to prevent siloed efforts.
- Systematically measuring the real impact of AI: clear, comparable KPIs reviewed by leadership, beyond initial enthusiasm.
- Investing in AI literacy across all levels—not just in technical roles.
The “era of scale” demands more than headlines about adoption; it requires embedding AI into the company’s operational DNA, always mindful that neglecting trust can cause the entire structure to collapse from the ground up.
FAQs on the CIO Study and the New Era of Scaled AI
What does it really mean that AI adoption has grown by 282%?
The study compares 2024 to 2025. A year ago, only 11% of surveyed organizations had fully embedded AI in critical processes. Today, that number has risen to 42%. This nearly fourfold increase in organizations with full deployments explains the 282% growth figure.
What is agentic AI in a business context?
Agentic AI goes beyond typical “chatbots that answer questions.” It consists of systems capable of executing end-to-end business tasks: consulting multiple internal sources, making decisions based on rules and context, updating records, triggering workflows, and in some cases, coordinating with other agents. In practice, they act as autonomous assistants specialized in areas like customer service, finance, marketing, or internal support.
Why has customer service become the primary testing ground?
Because it combines three factors: high interaction volume, highly structured tasks, and clear success metrics (response time, first-contact resolution, customer satisfaction). These factors allow CIOs to demonstrate quick, measurable results: less time per case, increased agent productivity, and better end-user experience. Once validated in service, many companies replicate the approach elsewhere.
If data trust is the main obstacle, where should a company start to scale AI?
Experts typically recommend three initial steps: inventory and classify critical data (know what exists and where), strengthen access and security policies (who can see what and under what conditions), and define a clear governance model (roles, responsibilities, review processes). Building on this foundation makes large-scale AI deployment feasible without risking security, compliance, or decision quality.
via: salesforce