Salesforce’s Agentforce Numbers Put a New Meter on Enterprise Software

TL;DR: Salesforce reported fiscal first-quarter 2027 revenue of $11.1 billion on May 27 and said Agentforce annual recurring revenue reached $1.2 billion, but the more important number may be the 3.8 billion Agentic Work Units it says customers have already used. The business implication is simple: Salesforce is trying to move enterprise software from seat-count economics toward work-meter economics, and investors are still deciding whether that protects the CRM franchise or exposes it.

##What Salesforce Is Really Measuring

The clean earnings headline is easy to read and easy to overrate.

Salesforce said first-quarter revenue rose 13% year over year to $11.1 billion, with current remaining performance obligation up 14%. It also raised the midpoint of full-year fiscal 2027 revenue guidance to a range of $45.9 billion to $46.2 billion.

Those are mature-software numbers. Good numbers, but not mysterious ones.

The more revealing piece is the company’s new habit of counting AI usage in operating units: 3.8 billion Agentic Work Units across Agentforce and Slack, up 111% quarter over quarter, plus more than 28.6 trillion tokens processed to date.

That is not just marketing vocabulary. It is Salesforce trying to teach the market a new meter.

#Why the meter matters

Traditional enterprise software sells access. A company pays for seats, clouds, storage, support, and renewals.

Agentic software wants to sell outcomes that can be counted inside the workflow: cases handled, sales steps completed, records updated, emails drafted, tickets routed, approvals nudged, or customer questions answered before a human opens the screen.

That difference is the whole story.

##Why Investors Are Still Uneasy

Reuters reported that Salesforce’s second-quarter revenue forecast came in just below Wall Street estimates, while investors remain worried that AI could disrupt traditional software demand even as Salesforce pushes Agentforce harder.

That is the contradiction sitting under the stock.

On one side, Salesforce says combined Agentforce and Data 360 ARR reached $3.4 billion, including $1.2 billion from Agentforce. On the other side, investors are asking whether AI agents expand the Salesforce bill or make some existing software seats look less necessary.

Both can be true for a while.

A support team might buy Agentforce because it wants to answer more customer questions with fewer manual touches. That creates a new AI line item. It can also make the old license conversation colder: if the workflow is doing more of the work, how many humans need the same expensive seat bundle?

That is why the casual “AI is good for Salesforce” take is too simple.

##Where The Real Business Test Happens

Picture a revenue-operations desk on renewal day.

There is a CRM dashboard open, a queue of stale leads, a support backlog, and a CFO asking why the software bill keeps climbing. The old answer was mostly defensive: the company needs the system of record because sales, service, and marketing all live there.

The new answer has to be more specific.

Salesforce has to show that the platform is not just where employees log activity, but where automated work is measured, governed, and billed.

That is a tougher sale in some ways. It invites a more practical buyer question: what did the agent actually do this month?

The answer cannot stay at demo level forever. It has to survive inside procurement, security review, and budget planning.

#The workflow ledger is the product

Salesforce’s May 1 reporting update said it is shifting fiscal 2027 revenue disclosure into two primary categories: Agentforce Apps and Data 360, Platform & Other. That accounting move is worth more attention than another product slogan.

It tells customers and investors where Salesforce wants the argument to move:

• away from isolated CRM modules;
• toward a data-and-agent platform;
• into measurable work units that can justify spending after headcount growth slows.

That is the right strategic direction. It is also less forgiving.

##Who Benefits If The Model Works

The first winner would be Salesforce itself, because usage-based work units could protect pricing power in a world where per-seat SaaS feels exposed.

The second winner would be large customers with messy workflows. Banks, insurers, retailers, software companies, and healthcare administrators already have too many handoffs between sales, service, finance, and support. If an agent can remove a repetitive handoff and leave an audit trail, the buyer can defend the spend.

The third winner may be the CFO who finally gets a cleaner map of software value.

But there is a catch. Usage meters cut both ways.

If Agentforce proves that one automated workflow can replace a pile of manual clicks, customers may become more disciplined about the rest of the stack. They may pay more for AI tasks and push harder on unused seats, overlapping tools, and legacy modules.

That is not a disaster for Salesforce. It is a business-model reset.

##What To Watch Next

The useful question is not whether Salesforce can mention AI more often. It can.

The useful question is whether Agentic Work Units become a number customers trust when they renew. If customers treat those work units like real labor saved, Salesforce gets a new growth meter. If they treat them like another vendor-defined activity metric, the company still has a strong CRM business, but not a clean AI re-rating.

The $25 billion accelerated share repurchase also matters here. Salesforce said the ASR delivered 103 million shares upfront and expects final settlement in the third quarter of fiscal 2027. Buybacks can support per-share earnings while the market waits for the AI model to prove itself.

That buys time. It does not answer the core question.

Salesforce is no longer only selling a place where work is recorded. It is trying to sell a meter for work itself. The uncomfortable part is that once customers can measure the work, they can also measure what they no longer need.

##FAQ

#What did Salesforce report for fiscal Q1 2027?

Salesforce reported $11.1 billion of revenue for the quarter ended April 30, 2026, up 13% year over year. It also reported $33.6 billion of current remaining performance obligation and raised full-year fiscal 2027 revenue guidance to $45.9 billion to $46.2 billion.

#Why do Agentic Work Units matter?

Agentic Work Units are Salesforce’s attempt to quantify work performed by AI agents across Agentforce and Slack. If customers accept the metric, Salesforce can make AI spending feel closer to measurable workflow output than another seat-based software license.

#What is the main risk for Salesforce investors?

The main risk is that Agentforce adds usage revenue but also pressures the old seat-based SaaS model. AI can expand Salesforce’s platform value while forcing customers to audit how much traditional software access they still need.