Salesforce Experiments with Multiple AI Pricing Models
In the last 18 months, Salesforce has utilized at least three different pricing models for its Agentforce AI product line. The company has shifted from per-conversation pricing to per-action Flex Credits and is now using per-user licenses. This rapid iteration suggests a broader SaaS trend of experimenting with pricing to align with evolving AI value propositions and buyer needs.
- The initial $2 per conversation model, launched in late 2024, caused significant customer confusion over what defined a "conversation," leading to slow adoption with only about 5,000 deals made in the first five months out of a customer base of over 150,000. - In May 2025, Salesforce introduced Flex Credits to align cost with activity, charging $0.10 per "action," such as updating a record or answering a query. These are sold in packs of 100,000 credits for $500. - To address enterprise buyers' need for budget predictability, Salesforce later added fixed per-user licenses. These include add-ons for Sales and Service Cloud at $125 per user/month and more comprehensive "Agentforce 1 Editions" starting at $550 per user/month. - Salesforce also introduced an "Agentic Enterprise License Agreement" (AELA), a flat-fee, multi-year deal for unlimited use of consumption products like Agentforce. This strategic move positions AI as a core productivity asset rather than a metered utility, with Salesforce indicating a willingness to lose money on some initial deals to drive adoption. - This pricing experimentation reflects a broader shift in the AI-native software market away from traditional seat-based models. Unlike most SaaS, AI products incur significant marginal costs for computation (inference), making usage-based or outcome-based pricing a better fit for value alignment. - The entire B2B software industry is in a state of pricing flux; in 2025, the top 500 B2B and AI companies made an average of 3.6 pricing changes each, and the adoption of credit-based models grew 126% year-over-year. - This trend extends to competitors, with companies like Intercom offering a "$0.99 per resolution" model and Zendesk moving to outcome-based pricing starting at $1.50 per automated resolution, highlighting the industry-wide search for a pricing standard that reflects AI's value. - A Salesforce CIO AI Trends study found that 90% of CIOs believe managing AI costs limits their ability to maximize its value, a key factor driving the creation of multiple pricing options to suit different procurement needs.
