About a week ago at TrailblazerDX, Salesforce’s developer conference, Salesforce announced a headless architecture that puts agents in the driver’s seat of applications and how people interact with them.
After reviewing the announcement, I was left with many questions — questions I’m hearing from CIO friends as well: Should CIOs rethink their approach to their current application stack? Should purchase plans still include seat-based subscriptions? How does Salesforce intend to win and differentiate as apps increasingly become databases orchestrated by agents automating tasks on behalf of humans? And how does Salesforce prevent market disintermediation in an open agentic world?
With those questions in hand, I reached out to a senior Salesforce leader for a direct conversation. I was fortunate to get an hour with John Kucera, senior vice president of product management, who answered each question with impressive depth and clarity.
Moving from systems of record to systems of engagement
A few years ago, author and venture capitalist Geoffrey Moore announced that enterprise software was shifting from systems of record to systems of engagement — that is, from applications that store and process data to those that enable interaction and experience around that data. Moore clearly had a great crystal ball — but agentic AI is changing things even further.
My first question was on behalf of my son James, a sales manager: Is Salesforce’s graphical user interface going away?
Kucera responded that there will always be some form of a graphical UI. However, he added, “we believe that we are entering an agent-first world,” and for that reason, “we are going to drive an agent-centric experience.”
Differentiating in a headless world
Kucera framed Salesforce’s open approach as a continuation of its heritage, not a departure. “We have always been open and we will continue to be open. Being so is powerful. Salesforce is the number one API out there — we discovered this when we bought MuleSoft.” Having worked at Boomi, I can confirm just how central connectivity to Salesforce is as an API.
When asked whether Salesforce risks becoming mere middleware or just a system of record, Kucera pushed back. Yes, Salesforce is a great system of record, but it is more than that. He implied that Salesforce’s domain knowledge will remain a differentiator in an agentic world, so long as the company stays open in how it operates. “We are evolving this way because we are a customer company, and customers are going to want choice and the openness that we provide.”
At that point, he slipped in a line I’ve heard from other vendors defending their core franchise: “It is hard to vibe-code our core asset.” He followed it with, “It is going to be hard to be as good as we are going to be at agents.” We’ll see.
How CIOs are rethinking the application stack
Many CIOs are slowing the extension of multiyear SaaS contracts, with some shifting to one-year terms. Against that backdrop, I asked whether this is changing how they manage their application portfolios. Kucera didn’t dodge the question — he acknowledged that change is already underway. As agents take on more work, the traditional per-seat subscription model makes less sense. In response, he said Salesforce is rethinking how it prices and monetizes its offerings.
He emphasized the need for flexibility. While a transition is clearly coming, Kucera suggested it may not be as disruptive as the earlier shift from licenses to subscriptions. That said, he acknowledged the risk of near-term revenue pressure during the transition. For now, his guidance to CIOs is pragmatic: continue purchasing subscriptions while the market evolves.
How will operationalizing agents affect work?
Kucera is clear that a human transition is already underway. In core CRM areas like support, work is shifting up the stack — humans are moving from Tier 1 to Tier 2 and Tier 3, while agents take on more frontline activity. He also argues that agents are beginning to reshape the sales process itself. Given that, I asked what the path forward looks like. His framing: a progression from steps, to tasks, to processes, and ultimately to broader business capabilities.
He put it this way: early generative AI use cases focused on automating individual steps — simple actions many people now perform with an LLM. That quickly evolved into tasks, such as retrieving pricing data or resetting passwords. Now, the focus is on processes — multi-step, multi-person workflows like mortgages or insurance claims. The next stage, he suggests, is coordinating multiple processes into larger, domain-level capabilities. This is what McKinsey is describing in Rewired Version 2 — and should be scary to companies like Appian.
While Salesforce has long been a leader in low-code platforms, Kucera believes the company is ready to move into these larger process domains. He describes himself as “a process guy” and sees this as a natural evolution. Today, processes are often framed as either customer- or employee-centric, but the real opportunity lies in orchestrating across them.
That raises the obvious question: will Salesforce expand into end-to-end processes like order-to-cash that cut across traditional ERP boundaries dominated by SAP and Oracle? Kucera’s answer was direct — yes.
With their MCP [Model Context protocol] server approach, he believes they can address these cross-system challenges and enable customers to take action across systems, not just within them.
What about Anthropic and Open AI?
When I asked whether Salesforce was worried about Anthropic and OpenAI, Kucera was confident.
“Anthropic is many things to us — a customer, a partner, a benefactor, a hosted solution, and yes, for certain things, a competitor,” he said. He was unwilling, however, to predict much further out than a few years.
So how does Salesforce intend to win as apps increasingly become databases run by agents automating tasks on behalf of humans? Kucera stressed that one size does not fit all, and that openness to the broader ecosystem is essential.
On the question of market disintermediation, he pointed to Agentforce’s early traction as a reason not to worry — at least for now. But he was candid, channeling something of a Doc Brown moment from Back to the Future: how software morphs and changes in the future is unwritten.
He said that vibe-coding on its own won’t reshape enterprise software, and also stressed that it is still early days. From there, he leaned into Salesforce’s current position. Agentforce is already moving beyond pilots — it is, he said, the fastest-growing product platform in Salesforce history.
“We are going to win by continuing to disrupt ourselves. To win, we have to innovate faster and make adoption possible for more of our customers.” The edge, in his telling, lies in being enterprise-grade, enabling successful adoption at scale, and helping managers transition into agent managers — a capability he sees as foundational to building what he called truly agentic companies.
What does this mean to CIOs?
CIOs are right to be wary. Salesforce, like many vendors, is in the middle of a significant transition. To its credit, it is disrupting its own model and moving toward the openness the market increasingly expects. At the same time, trends like “vibe coding” are unlikely to replace the core code and architecture that underpin its franchise.
The open question is what this means for Salesforce’s long-term position. As Rita McGrath has argued, companies at inflection points must “see around corners.” Salesforce appears to be doing just that. But capital markets have reason to be cautious. As McGrath notes, digital technologies can fundamentally reshape how value chains are structured and managed.
That creates a paradox: Salesforce can execute well — embracing agents, openness, and new pricing models — and still find its role in the ecosystem shifting as the rules of the game change. It’s simply too early to call. Expect this to remain an evolving story worth close attention in the months ahead.
