This week, we look at another ETR Insights Interview. In this one, the CIO of a midsized business services firm shares how a once-optimistic 2025 IT budget has taken a sharp turn, flattening out and, in some areas, shrinking. The shift has led to delayed refresh cycles, paused initiatives, and tighter oversight on managed services spending. What remains on the priority list? Digital transformation, cloud, and security.

The company is consolidating onto Salesforce. As long-time users of Slack and Tableau that picked up Marketing Cloud and Service Cloud along the way and, most recently, piloted Agentforce, they are finding that tighter native integration, per-seat bundle discounts, and early success in automating frontline customer service interactions justify the move. Overall company spend on AI, including Microsoft Copilot, is expected to jump in 2026; Agentforce-driven automation may handle the majority of customer service volume within 12 months. Read on for more on Salesforce’s aggressive pricing incentives, this CIO’s framework for AI return on investment, and plans to roll Agentforce into additional languages and regions.

Vendors Mentioned: Amazon (AWS) / Anthropic / Databricks / Google (Gemini) / HubSpot / Microsoft (Azure, Azure Data Lake, Copilot, Microsoft 365, Power BI) / Salesforce (Agentforce, Einstein, Marketing Cloud, Service Cloud, Slack, Tableau)

The Show Must Go On: Macro Uncertainty Curtails, but Doesn’t Halt, IT Spending Growth

Despite earlier plans to increase technology spend by as much as 7%, our guest now expects no growth at all and is bracing for outright cuts. A downturn in the broader market has forced the organization to weigh tough trade-offs. “We may even contract even more midstream, trying to squeeze a few dollars out of cloud optimization, or license covering, things like that across the board. But principally, it’s delaying projects, delaying refreshes, maybe pausing certain key initiatives.” Complicating matters, this reversal happened at breakneck speed, spurred by the evolving geopolitical and financial-market environments. “It’s not like a 2009 situation where suddenly the brakes hit and everything just… it’s not that. It’s more like, we need to really draw this in, and we need to do it now. How do we best achieve that?”

Even as it trims overall spending, our guest’s organization is doubling down on core technology initiatives, such as digital transformation and cloud. “Security remains a high priority. That’s probably one area that’s going to survive budget cuts, but because it’s not directly being cut, we have to find other places to make reductions.” To offset those protected allocations, the company is slashing outlays for externally managed services and squeezing savings from everyday IT operations. “We’ve even done some smaller things like extending the refresh life cycle on things like laptops, delaying certain enterprise software purchases, and rethinking the number of seats that are part of renewals and/or net new purchases.”

ETR Data: Faced with difficult decisions, our guest’s organization is decreasing their IT budget using similar methods to their peers. In the APR25 Macro Views Survey, reducing staffing and consulting costs, delaying new projects, consolidating vendors, and reducing hardware spend were the most common methods for achieving a budget decrease.

Agentforce is a Natural Choice for those Within the Ecosystem, but Data Must be in Order

Agentforce is logging faster-than-expected uptake, but only where Salesforce is already deeply embedded. Our guest explained how a homogeneous vendor stack can vastly simplify AI deployments like Agentforce, with tighter integrations and bundle discounts. He says that enterprises with diverse data infrastructures should anticipate longer timelines, higher costs, and potentially diminished early ROI from AI-driven solutions. “If you think you’re going to take this on yourself, especially with diverse data sets – it’s probably going to be pretty painful. But if you’re Salesforce-oriented, and you work directly with Salesforce or an experienced partner, it can be done pretty quickly and easily, at least to start.”

Our guest cautions over-excitement and reiterates that the value of technology is really unlocked when data is robust on the platform, hesitating to classify Agentforce as a singular, groundbreaking solution. “Is this just a rewrapping of Einstein? Yeah, it probably is. But Einstein adoption, prior to now, Agentforce, was pretty low. So, I think there were a couple of things that happened the last few months…the rebranding, the seed money.” Outside that captive base of Salesforce bundle customers, he suggests the product holds little edge over generic AI offerings like ChatGPT.

The organization also maintains a significant investment in Microsoft’s ecosystem, leveraging Azure, M365, and evaluating Power BI against Tableau. At this point in the organization’s AI journey, our guest sees better immediate synergy and pricing leverage within Salesforce’s portfolio. Nonetheless, they do use Microsoft Copilot as a “general-purpose” AI solution (vs. specialized solutions like Agentforce).

Our guest notes that Salesforce is cutting unusually sharp deals for customers willing to trade out competing platforms, pricing flexibility he says he hasn’t seen in half a decade. He posits that these steep discounts are the driving factor behind (at least some of) Agentforce’s early momentum, though he suggests that the newfound pricing aggressiveness is likely a short-term, strategic push rather than long term. In 2024, our guest’s organization spent virtually nothing on AI, but the category has since climbed into the third quartile of technology spending and is projected to reach the second quartile by 2026, matching areas such as security and cloud observability. “Where we’re going to find the money for that, it’s going to have to come at the expense of other things in the budget. Because we’re not going to just net increase the budget to cover this.” Given their existing alignment with Salesforce, the organization decided to take the bait. “There’s a window of opportunity open right now to partner with Salesforce, to get this foothold with Agentforce, and to do it at very low cost. That opportunity is going to close soon, and then once it has, you’re going to be chasing your competitors.”.

ETR Data: In ETR’s NOV24 AI Product Series, Salesforce Agentforce registered a 43% attach rate with existing customers, which is relatively middling, though the product register’s survey-leading citation rates.

The company’s early applications for Agentforce are “predominantly to reduce or eliminate redundant, tedious work, or to reduce payroll, especially around frontline kinds of roles, namely customer support. I completed our pilot Agentforce initiative late last year, and now we’re kind of expanding and building upon that.” Last year, they began a pilot within Service Coud, which has already proven fruitful. “It’s self-serve, self-help, FAQ-driven, and things of that nature, with a kind of a two-part goal. One, to improve in customer satisfaction by getting them quick and easy resolutions to frontline issues, and two, to drive down customer support operational costs, namely through people. It’s early, but we have seen both of those things play out the way we’d hoped… so far.” The company is readying a broader push: more complex tickets, additional languages and currencies, and wider geographic coverage.

At the end of the day, our guest says that end-customers interacting with automated solutions like Agentforce appear largely indifferent to the technology powering their experience. Rather than being concerned about engaging with an AI-driven solution, customers primarily care about outcomes: “Our overall satisfaction scores are up. I don’t think our customers are thinking at all about how we’re providing support. They just know they’re getting resolutions quicker and better.” Customer-facing automation need not be overly transparent – adoption will ultimately hinge on performance, not novelty or sophistication.

While our guest remains cautious about predicting exact headcount reductions attributable to AI, he acknowledges that new investments in automation will certainly shape hiring decisions going forward. Specifically, Agentforce has allowed the company to reconsider filling customer support vacancies immediately: “We’re now in a position to start to say, maybe we don’t have to fill some of our customer support vacancies, or we can delay filling them to see how Agentforce plays out. Pre-Agentforce, we would not have had any of those options.” This reflects a broader market sentiment – AI’s labor impacts will likely manifest gradually, initially curtailing growth rather than abruptly cutting existing staff.

ROI measurement for AI investments at this firm unfolds in two stages. Initial success indicators revolve around improved customer satisfaction scores, driven largely by quicker problem resolution. However, our guest emphasizes the second, longer-term assessment stage as crucial: “Eventually, probably in a year’s time, we’ll start measuring the fully baked cost of Agentforce – no seed money, no ‘loss leader’ pricing – offset against payroll savings in customer support. Only then can we really measure how favorable or unfavorable our investment truly is.”