Best pricing model for AI? Work in progress, says Salesforce
Is that 'best' for customers or for shareholders? Any vendors that think they've got this 'all figured out is kidding themselves'
A senior Salesforce exec says users need to be flexible about AI pricing models while vendors determine which one works best.
Speaking at a conference hosted by investment bankers Jefferies, Bill Patterson, Salesforce executive veep for CRM applications, said any vendor who thinks they have it "all figured out is kidding themselves."
He said AI providers – including standalone model providers, cloud companies, and application vendors – were trying to figure out which model "works best for the use cases."
Last year, Salesforce mooted, and then introduced a pricing model based on a $2 charge per AI agent conversation. It later added a charging model based on the actions of AI agents.
In mid-May, Salesforce announced new Agentforce pricing, which includes both conversation and action-based pricing – whether that's updating customer records, automating complex workflows, or resolving cases. It also debuted a credit-based system to allow users to move between both of these models and user-based licensing.
- Look forward to having a Slack teammate, says Salesforce's Benioff
- Salesforce takeover of Informatica is on for $8 billion
- The future of LLMs is open source, Salesforce's Benioff says
- No new engineer hires this year as AI coding tools boost productivity, says Salesforce
Salesforce introduced Flex Credits to offer "customers unprecedented flexibility with a consumption-based model that aligns cost with business outcomes," the company claimed.
For example, customers can convert user licenses into Flex Credits, or Flex Credits into new user licenses, "exploring new, value-generating use cases."
Patterson said: "Not every use case is a conversational affair that agents can do, so you're going to see a broad spectrum – the hyperscalers monetize AI [with] tokens, the applications companies do it in either sort of conversations or actions; or even a third one – which you'll probably see more of from us – per-user-per-month offerings."
Some vendors, he added, were considering outcome-based pricing, which was an "interesting one, but is hard to scale, because sometimes the outcomes are debatable" between the contributions from the AI software and the person presumably in charge.
Whatever the model, CEO Marc Benioff has assured investors that the move to agentic AI will be "a very high margin opportunity" – for Salesforce. ®