Product Sense Pushup 7: Three Ways to Price AI

ChatGPT has a flat-price per user with freemium pricing model – free tier, $20 Pro and $100 Max. Each tier has usage thresholds for a user, once the threshold is reached the user is then downgraded to a lower model to reduce their usage. With this model, the cost of the heavy user falls on OpenAI. This model is charging for access to the technology.

Github Copilot charges $19 per head, which gives each user $19 worth of AI credits. Once this $19 is depleted, the user will pay for any additional usage. With this model, the cost of the heavy user falls on the customer. This model is charging per person.

Intercom Fin charges $0.99 per resolved conversation, which is also refunded if the resolution is reversed. Intercom ends up bearing the cost of expensive/arduous resolution cases for the same fixed $0.99 and also carries that same cost even if the resolution is not resolved. This model is charging for outcome.

For our product, Doorman, I think a pricing model like GitHub Copilot (charging per person) is better, as it puts the cost on the company. Companies are charged based on the difficulty/magnitude of the contexts that Doorman has to digest in order to give sound recommendations. This means that we are not incentivised to compromise the quality of recommendations for speed.

 

Avatar

About the author

Leave a Reply