The Third Commercial Transformation

The enterprise software industry is in the middle of its third major commercial transformation in twenty-five years, and most of the vendors going through it, my own employer included, are still in the early stages of working out what it will actually require. The first transformation moved software from perpetual licenses to subscription, which took roughly a decade to complete and forced a rebuild of nearly every commercial function in the business — pricing, sales comp, deployment motion, customer success, financial reporting, and the story told to investors. The second transformation introduced consumption pricing in infrastructure and data, led by a small group of companies — Amazon Web Services, Snowflake, Twilio, Databricks — that treated usage as a more honest proxy for value than seats had ever been. That second transition is still playing out, unevenly, across the rest of the software landscape. The third transformation is the arrival of agentic software — software that acts on behalf of the customer, autonomously, inside the customer's workflow — and it is the most difficult of the three because the commercial unit is genuinely unclear for the first time in the history of the industry.

I am not alone in reading it this way. IDC's FutureScape: Worldwide Agentic AI 2026 Predictions projects that by 2028, pure seat-based pricing will be obsolete, with 70 percent of software vendors refactoring their pricing strategies around new value metrics such as consumption, outcomes, or organizational capability.¹ Forrester's 2026 Enterprise Software predictions frame the shift in broader terms — from a user-centric design philosophy to a worker- and process-centric one, with enterprise applications evolving beyond the role of enabling employees toward accommodating an emergent digital workforce of AI agents.² These are not speculative calls about a distant future. They describe a structural movement already visible in vendor behavior today, and one that will accelerate substantially over the next twenty-four to thirty-six months.

The core difficulty of the third transformation is that a seat does not make sense when no human is touching the product at the moment of value creation. A consumption unit makes partial sense, but the cost structure behind that consumption is lumpy in ways the SaaS era never had to accommodate, because agents use model calls and tool calls and orchestration cycles and real-time data retrieval in proportions that vary wildly by use case and customer. An outcome-based price is the theoretically correct answer, but the instrumentation required to price on outcome — the shared observability between vendor and customer, the agreed-upon measurement frame, the contractual machinery to settle against results rather than against usage — does not yet exist in most enterprise relationships. Gartner had projected that by 2025, over thirty percent of enterprise SaaS solutions would incorporate outcome-based components, up from roughly fifteen percent in 2022,³ and while the direction of travel is clearly correct, the operational reality lags the forecast almost everywhere I look.

You can see the industry working through this in real time. Intercom has been pricing its Fin AI agent at ninety-nine cents per resolution, a genuinely outcome-based model that only functions because a customer service resolution is among the more cleanly measurable outcomes in enterprise software. Microsoft has been consolidating its specialized copilots into a single Microsoft 365 Copilot, reinforcing a simple per-user model that encourages broad adoption at the cost of more granular value attribution. Salesforce originally introduced Agentforce at two dollars per conversation, and then in late 2025 introduced the Agentic Enterprise License Agreement — AELA — which offers unlimited use of Agentforce, Data 360, and MuleSoft for a fixed fee over two or three years, with CRO Miguel Milano publicly accepting that the company is willing to lose money on some deals in exchange for customers who compound value over the duration of the relationship.⁴ Three vendors, three very different commercial answers to the same underlying problem, within a single twelve-month window. That is not a stable market. It is a market in active redefinition, and the companies that redefine it deliberately will shape how the rest of the industry eventually settles.

I have watched a structurally similar pattern play out twice from the inside now, which is part of why I find the current moment so familiar. Several years ago at Salesforce, when I was running product marketing for Data & Identity, I watched the company pivot from a portfolio of point clouds into the unified Customer 360 narrative. The category insight was correct and it still is: data is the substrate that makes the portfolio more valuable than the sum of its parts, and the competitive moat is built in the unified customer profile rather than in any single application. What I learned from that period is that a transformation of that magnitude takes longer to propagate through the commercial machinery than the category narrative implies, and that the narrative and the commercial motion have to be deliberately kept in conversation with each other — because if they drift, customers feel the drift before analysts do, and the feeling is one of quiet confusion rather than articulated complaint. The point is not that any particular vendor handled this imperfectly. The point is that every vendor going through a category transformation of this magnitude faces the same choreography problem, and the ones that invest early in closing the gap between story and contract are the ones that extend their lead.

I am watching a structurally similar dynamic from my current seat at SAP, where I work as the AI lead within the Data & Analytics team. The category narrative across the enterprise software industry over the last eighteen months has organized itself around the agentic enterprise — the proposition that the next generation of enterprise software is not a collection of applications but an orchestrated set of agents operating on unified, governed business data. That narrative is strong, and I think it will prove substantially correct across the category. The broader industry challenge, which SAP shares with every other major vendor, is that the commercial machinery across the enterprise software landscape was built for the first two transformations and has not yet been rebuilt for the third. Procurement organizations on the buy side do not know how to buy agents, which lengthens sales cycles that should be shortening. Pricing pages across the industry use vocabulary borrowed from the SaaS era because the vocabulary for the agentic era has not yet been agreed. And marketing organizations everywhere, mine included, are describing a transformation while the commercial motion around them still speaks in SKUs. This is not a failure of any individual vendor. It is a description of the category in the middle of a structural shift, and the vendors that address the narrative-to-commercial gap earliest will compound advantage the rest of the industry will struggle to close.

The clearest current illustration of the gap is the packaging of conversational marketing capabilities across the category. Every major marketing software vendor has spent the last year telling marketers that the category is shifting from one-way campaign broadcasting to two-way conversational engagement, and that the brand of the future will converse with its customers rather than blast at them. That diagnosis is almost certainly correct, and it is the most important message in the category. But across the market, two-way conversational capabilities remain largely packaged as premium tier upgrades, frequently with cross-cloud dependencies, which places the capability at the center of the category narrative behind a commercial wall. This is a defensible packaging choice for any individual vendor, because conversational infrastructure has real incremental cost and delivers real incremental value. From a category-leadership standpoint, though, it creates an uncomfortable pattern: the marketing conversation and the sales conversation describe two different products. A category leader cannot indefinitely sell a story whose central claim lives behind an upsell, because customers are perceptive enough to notice when the narrative and the contract stop rhyming.

The resolution of that tension will not come from a cleaner pricing page or a better SKU rationalization. It will come from a reconception of what the product fundamentally is, which requires the marketing organization to operate adjacent to product management in a way most marketing organizations are not currently structured for. Forrester's recent analysis of Salesforce's AELA move makes this point with unusual clarity — AELA is not a pricing innovation so much as a reframing of AI agents as productive assets rather than metered utilities,⁴ which implies that the buyer conversation has to shift from usage questions to capital-allocation questions: what economic output will this generate, what is the ROI and IRR, what is the useful life of the agent. A vendor cannot ask its customers to have that conversation while its own commercial motion still describes the product in SaaS-era vocabulary. The story and the commercial motion have to be architected together, which in turn implies a degree of cross-functional authority that most marketing organizations currently lack, and that the next generation of marketing leaders will need to claim deliberately.

There is a second structural problem that compounds the first, and that I have been writing about at length in my forthcoming book, Agent Driven, as well as in the introductory piece on the AMP I published here earlier this year.⁵ The problem is that the buyer of agentic software is itself in the process of becoming partly agentic. The CMO evaluating whether to buy a marketing platform from Salesforce or Adobe or Braze or any of the AI-native entrants is, today, running early procurement agents that read vendor websites, compare feature coverage, pull third-party review data, interrogate pricing and documentation, and produce structured recommendations for human review. This is not speculative and it is not a 2030 problem. It is in pilot form across the Fortune 500 right now. The marketing function responsible for making a software product findable, comparable, and selectable by such an agent does not have a name yet, does not have a budget line, and does not have a playbook. It is a category of marketing work that product marketing has not claimed, that brand has not claimed, that content has not claimed, and that demand generation has certainly not claimed, because the conventional demand funnel assumes a human is reading your page — and in the emerging model, the human may never read your page at all.

In the book, and in the AMP article, I refer to the asset that addresses this problem as the Agentic Management Platform, the structured, machine-readable, continuously maintained representation of what the brand does, who it serves, how it competes, and why it should be selected. The AMP is not the website. It is not the sales deck. It is the substrate that buyer agents consume when deciding whether to include a vendor on the shortlist, and it will become a material portion of the marketing discipline over the next three to five years. The vendors that build a serious AMP strategy early will benefit from a form of compounding advantage analogous to what search engine optimization produced for the early movers in the mid-2000s. The vendors that treat it as a technical side project, or that fail to claim it as a marketing responsibility at all, will spend the second half of the decade wondering why their pipeline degraded in ways their conventional analytics could not explain.

The job of the marketing software CMO in the agentic era is therefore not to run a better campaign engine, or to modernize the demand funnel, or to relaunch the existing portfolio under a new brand architecture. The job is to lead the organization through a commercial transformation of the same magnitude as the shift from license to SaaS, on a compressed timeline, with a less-settled endpoint, and while the existing business continues to generate the quarterly revenue that funds the transition. That requires a specific combination of capabilities — the ability to architect a category narrative that can survive contact with the pricing page, the willingness to push the commercial model toward a version of value-based pricing the organization is not yet built to sell, and the discipline to address the buyer-side inversion before it becomes obvious to everyone else in the category. The vendors that fill that seat with the right leader will compound advantage for a decade. The vendors that fill it with someone who treats agentic as a campaign upgrade will spend the same decade wondering where their market went.