Salesforce (NYSE:CRM) faces a question that goes to the core of its growth story: can agentic AI translate into durable, recurring revenue rather than one-off pilots and headline demos?

The company has positioned its Agentforce platform — software agents built to handle service, sales, and support tasks autonomously — as its next major revenue layer beyond the core CRM subscription business. The underlying commercial bet is straightforward: customers pay for outcomes and usage, not seats. If agents resolve conversations or complete workflows, Salesforce bills for those actions.

That model matters for how investors should read the numbers. Seat-based SaaS growth has slowed across the enterprise software sector, and Salesforce is not immune. Agentic AI offers a different monetization path: revenue tied to consumption of agent-driven work. The question raised in Kalkine Media's analysis is whether that consumption is material yet — and whether it compounds into real annual recurring revenue.

The skeptical reading is worth stating plainly. Enterprise AI deployments frequently stall between pilot and production. A customer experimenting with agents in a sandbox generates press but little billable usage. For agentic AI to count as recurring revenue, customers must run agents in production, at volume, month after month — and keep paying when contracts renew.

The optimistic reading rests on Salesforce's installed base. The company sells to a large share of the enterprise CRM market, which shortens the distribution path for a new product tier. Cross-selling an agent layer into existing service and sales clouds is structurally easier than acquiring net-new customers. Margin dynamics also differ: usage-based agent billing can expand revenue within an account without adding headcount on either side of the contract.

What separates the two readings is disclosure. Recurring revenue from Agentforce would show up in specific line items — paid agent consumption, contract values attached to AI tiers, retention figures for customers that adopted early. Investors looking to interrogate the claim should watch whether Salesforce reports these figures with methodology and sample sizes, or whether agentic AI remains an asserted growth driver without a measured base underneath it.

There is also a competitive dimension. Every major enterprise software vendor — Microsoft, ServiceNow, Oracle among them — is shipping agents into the same customer base. First-mover pilots rarely decide these contests; renewal economics do. Salesforce's argument depends on agents performing well enough in production that customers build workflows around them and face switching costs at renewal.

For the sales teams selling inside this ecosystem, the shift carries practical weight. If agent consumption becomes the billing unit, account plans built around seat counts and user adoption lose precision. Pipeline value starts to track automated work delivered, not licenses issued — a change that alters how sellers forecast, how compensation plans get written, and how expansion within an account gets measured.

The title of Kalkine Media's piece frames this as an open question, and that framing is accurate. Salesforce has the distribution and the product surface to make agentic AI a genuine revenue line. Whether the usage data, once reported, shows compounding consumption or a long tail of stalled pilots will determine if the bet pays off. The next several quarters of paid-usage disclosure should give the market its answer.