As a substitute of charging based on the quantity of AI chatbots used, a brand new charging method has emerged that charges only when tasks are accomplished. As firms try to cut back the price burden of AI, it stays to be seen whether this method will grow to be a serious trend in AI software price battle.
The Information reported on the sixth (local time) that AI software company Zendesk introduced an ‘outcome-based pricing’ method.
Performance-based billing is different from traditional software-as-a-service (SaaS). Most AI software firms charge based on the variety of users or usage. SaaS has been the usual within the software industry for the past 20 years.
Nevertheless, Zendesk combines a usage-based model with a performance-based model in its pricing model, which charges businesses for AI resolutions per user. Which means as an alternative of charging businesses for the way often they use the service, they only get charged when the chatbot routinely completes a task without human intervention.
It’s also known that increasingly more software firms are adopting this method. Along with Zendesk, Intercom and Foresort also charge only when AI functions work well routinely. Intercom charges 99 cents (about 1,300 won) per resolution, and Foresort requires users to buy a certain variety of AI resolutions upfront.
“In an industry that’s rapidly becoming more automated, traditional software pricing structures have gotten less relevant,” said Nikhil Sane, senior vice chairman of revenue acceleration at Zendesk. “Simply because you employ a service doesn’t mean you’re getting value.”
As businesses increasingly use AI, there are growing voices saying this approach will grow to be the norm, as many users are already wary of pricy AI features.
Nevertheless, there are also those that indicate that the performance-based payment system continues to be in its early stages and that it’s unknown what the outcomes shall be. For service providers, the shortcoming to predict revenue is a serious weakness. Specifically, the issue becomes greater when AI doesn’t work as intended.
There was also an opinion that it’s difficult to place a selected price on the worth of AI solutions. It was explained that even among the many same users, and even between departments of the identical company, there shall be differences in the way in which they give thought to the worth they get from AI solutions.
Nevertheless, there may be a powerful opinion that increasing using AI is advantageous to competition even when it risks a short-term decrease in sales.
Within the US, there was a recent increase in cases of custom models being built by fine-tuning open source attributable to cost issues, and this phenomenon is usually a fatal blow to software providers.
OpenAI also released a low-cost model like ‘GPT-4o Mini’, so the burden on firms that provide software using it has been reduced. Likewise, firms that directly operate large language models (LLMs) are also lowering their fees.
“For those who can market AI cost savings as a selling point, it might ultimately drive huge revenue, unlike traditional enterprise apps,” said Joan Chan, a partner at Foundation Capital. “It is a huge market expansion opportunity.”
Scott Woody, CEO of Metronome, also identified that “if some firms begin to lower their prices, their competitors can have no alternative but to quickly follow suit.”
Meanwhile, in China, where greater than 100 AI models have been in the marketplace, AI price battle has already begun because the starting of this 12 months.
In May, Alibaba slashed the costs of its AI services by as much as 97%, and Baidu responded by offering free services. China’s No. 1 chatbot, ByteDance’s Dubao, also slashed its prices.
Reporter Park Chan cpark@aitimes.com
