At F3, staying current on all aspects of go-to-market strategy means talking directly to the people with expertise in niche areas. In that effort, I sat down over Zoom with Kaytlin (Kayt) Parrella of 28 North Consulting and Raul Acosta of Klient to discuss AI in Salesforce, RevOps, professional services automation, and the broader direction of the ecosystem. Between them, Kayt and Raul attended a collective five Agentforce World Tour events this year.
The Agentforce World Tour is Salesforce’s series of one-day regional events built around AI, autonomous agents, and the shift toward what Salesforce calls the agentic enterprise. Each stop brings together customers, partners, and Salesforce leadership to discuss where the platform is headed and how organizations are putting it to use. Through the conversation, Kayt and Raul described how the messaging at these events has shifted from stop to stop and month to month, tracking closely with how the market itself is evolving.
What stood out most, however, was not the messaging shift, which as a messaging nerd, would normally catch my interest. It was what sat underneath it. Salesforce has rewritten Agentforce pricing three times in the past 18 months, a detail worth sitting with on its own. And beneath that, Kayt and Raul each pointed to a specific gap in the current conversation, one that neither Salesforce nor most of the ecosystem is addressing directly.
Pricing Instability as a Market Signal
Raul, who worked at Salesforce before joining Klient, pointed to a concrete data point. Agentforce pricing has changed three times in roughly 18 months. It started as a seat-based SaaS model, shifted to consumption-based pricing, and has now settled into a hybrid of consumption and outcome-based pay-per-resolution, announced in June.
His assessment was direct. In his time at Salesforce, pricing changes at that pace, on a single product, typically only happen in response to significant buyer resistance. The implication is that the market has not yet settled on how AI capability should be priced or valued, and Salesforce is adjusting in real time to figure out what customers will actually accept.
That instability is not incidental. It reflects a deeper problem neither Salesforce nor most partners have solved, which is that organizations still lack a reliable way to forecast what AI adoption will cost once it moves past a pilot.
Two Gaps, Two Vantage Points
Asked what topics are not getting enough attention across these events, Kayt and Raul identified two distinct gaps, arrived at from different sides of the business.
Kayt’s observation centers on organizational readiness. Companies have absorbed the message that AI adoption is no longer optional. What they have not absorbed is that becoming ready for it carries a real cost. Few organizations are budgeting for the work required to get RevOps functions in order before layering AI on top. The assumption tends to be that readiness happens automatically during implementation. In practice, a new process can be documented in an afternoon, but changing how people actually work takes weeks or months, and that timeline is rarely planned for or funded.
Raul’s observation centers on cost visibility. Salesforce can demonstrate a strong Agentforce use case, but no one is demonstrating a credit consumption forecast. That gap between a capability demo and a realistic cost projection is significant enough that Raul is now seeing large enterprises, including sophisticated technology companies, pull back on unrestricted Agentforce credit consumption once the financial impact becomes visible. At Klient, this has led to building cost modeling tools for customers and recommending a crawl-walk-run approach so organizations understand consumption patterns before scaling AI use across departments.
Taken together, these two observations point to a shared root cause. Organizations struggling with AI adoption are not struggling because the technology underperforms. They are struggling because the foundational work is not being planned for with the same rigor as the technology itself. On one side that means employee resource allocation and process change. On the other, it means cost forecasting.
The Build-Versus-Buy Trap
A related pattern surfaced in the conversation, illustrating the same gap at a smaller scale. Kayt described a recurring client question about why they shouldn’t simply build their own tools rather than adopt a platform like Klient’s PSA solution. The logic is understandable on its face. In practice, organizations that take this route end up with a tool built around how they operate today, without the accumulated best practices embedded in a purpose-built product. They then own the ongoing burden of supporting it, with no vendor partner continuing to improve it over time, which typically results in an isolated system that becomes a long-term liability.
This is the same gap Raul and Kayt each flagged, expressed as a build-versus-buy decision instead of a budgeting decision. Skipping the foundational work, whether that is process design or product maturity, tends to cost more later than it saves upfront.
What to Expect Going Forward
Both Kayt and Raul were asked to look ahead a year, and their answers converged from different directions.
Kayt expects the current focus on customer service use cases, agents answering routine questions and routing cases, to expand into sales and marketing as organizations grow more comfortable with agentic workflows. Raul is watching for a shift at this year’s Dreamforce specifically, away from polished demonstrations and toward evidence. He expects the emphasis to move from the art of the possible to adoption data, revenue impact, and outcomes tied to a P&L line rather than a stage demo.
Neither prediction assumes a technological breakthrough. The technology has already reached sufficient maturity, and that the market is becoming less tolerant of unproven claims and more focused on demonstrated results.
The practical takeaway is that the most successful organization over the next year will be the ones that treat budgeting, process change, and cost forecasting with the same seriousness they applied to the technology itself.
You can read the full interview with Kayt and Raul here.
Back to Blog