|
A wave of AI pricing changes is keeping customers on their toes. Anthropic, Salesforce, ServiceNow and Workday have shifted to charging for AI based on how much customers use, rather than a flat subscription fee, for instance. Now Adobe is taking steps to join AI startup Sierra and other firms that want to go a step further, charging customers based on how well their AI tools work, such as when they help businesses complete tasks, according to Adobe president Anil Chakravarthy. Adobe will use the pricing policy for its newly rebranded suite of AI products released this week, CX Enterprise, which includes AI agents that can help customers complete “whole jobs.” For example, Chakravarthy said CX Enterprise can pull data from Adobe apps as well as non-Adobe software such as Amazon Web Services databases to figure out why a hotel company is experiencing low bookings for hotel reservations in the south of France. Adobe hasn’t disclosed specifics around the new pricing, but it will be based partly on the value of what the product produces, the executive said. For example, he said pricing could be tied to the number of ad campaigns that AI agents in the CX Enterprise complete for a travel or hospitality business. Previously, Adobe’s AI pricing was subscription-based or tied to consumption. Chakravarthy criticized the idea of charging businesses based just on AI usage, or how many tokens—pieces of data—the AI processes. He said the Photoshop maker will instead move toward outcome-based pricing. “Tokens don't equate to value,” he said. He said that some customers may not be using AI efficiently, which means they are overpaying. In those cases, suppliers are “just passing on the cost” of providing the AI tools without creating value for customers. (Chakravarthy didn’t name names, but Anthropic is the most prominent firm to go big into usage-based pricing, as my colleagues reported last week.) Adobe isn’t completely abandoning usage-based pricing. A spokesperson said Adobe would still use that approach, as well as subscription-based pricing, for some AI tools. Also, the new pricing tied to outcomes does not yet apply to the company’s existing AI tools for helping with tasks such as photo and video editing. It isn’t clear if the new pricing will be more or less expensive than subscription pricing for most customers. Adobe’s move seems similar to the AI pricing that Zendesk and Intercom, which automate customer service interactions, pioneered a couple of years ago. Brett Taylor’s AI startup Sierra, a rival to those firms, also charges customers based on how many tasks AI agents actually complete, such as selling a product or resolving a customer service issue. Salesforce, which also sells AI to automate customer service, said in February that it is beginning to measure how many tasks its AI agents complete for customers, rather than simply measuring how many tokens they process. It even coined a new metric, the Agentic Work Unit, an indication that the software firm could be thinking about pricing its AI tools based on that unit of measurement. I asked Chakravarthy how Adobe’s pricing change will affect the predictability and profitability of its customer contracts, as some companies might have to consume more tokens than others to generate similar business outcomes. “Some will have higher spikes in their utilization [of tokens],” he said, but “we will get efficiencies over time.” Like many software executives, he argues that prices for AI models will continue to fall over time, and customers will focus on measuring the value of their purchases. That’s a bet that could pay off: If running AI models gets cheaper, software firms that charge based on results, rather than usage, would be in a good position.
| 
