Salesforce wants to charge for AI outcomes, but first it needs to figure out how

Nothing demonstrates how deeply Fortune 500 companies depend on Salesforce quite like a global outage lasting more than seven hours.

The interruption struck during the second day of last week's Dreamforce conference, where the $40 billion-a-year SaaS vendor was pitching AIforce and a growing collection of AI products.

The question now is how the company will charge for them.

AIforce illustrates why the per-user licensing model is becoming harder to sustain.

Unveiled at Dreamforce, it uses Salesforce's Headless Toolkit to make the company's data, workflows, and business logic available through interfaces including Slack and Claude.

When AI agents and APIs perform the work instead of named human users occupying seats, charging per user becomes a less natural fit.

Earlier this month, Bill Patterson, executive veep and general manager of CRM applications, told an investor webinar that Salesforce was devising "a new pricing structure that really aligns to the benefits that customers realize from this new technology." Under one such arrangement, AI customer service agents would be priced according to the cases they resolve.

Outcome-based pricing will not suit every use case, however.

Patterson acknowledged that some agents operate across multiple disciplines, domains, and products, making it difficult to identify a single measurable result.

Salesforce is therefore also developing bundles and using Flex Credits to charge customers according to consumption.

The company has reason to look beyond traditional licensing.

License revenue is dragging on growth, while AI agents complicate an approach built around charging for human users.

"Licenses are weighing things down right now.

They're a headwind," Mike Spencer, Salesforce's deputy CFO and head of finance, told the Deutsche Bank 2026 Technology Conference last month.