Skip to main content

How to price your AI Agent

With Nevermined it is possible to monetize any AI Agent you create. But pricing it correctly is key to making it a successful business and can be challenging. To do this, you need to take several factors into consideration:

  1. What LLMs or models are you using?
  2. Is your Agent using any additional services or data?
  3. Is your Agent using a lot of infrastructure resources? In the Cloud? GPUs?
  4. What's the architecture of your Agent? Is it a simple model or a complex one? Is a prompt or query to your Agent cascading into multiple steps going across different models?

A complete analysis of these factors will help you to price your AI Agent correctly. After that analysis, you can continue with the following steps:

  1. Understand the average cost of the different models and services per query.
  2. Understand the fixed cost of running the Agent (for example, the monthly cost of the infrastructure).
  3. For regular interactions (simple/complex) with your Agent, analyze the execution plan. This means determining how many queries to the models are needed and how complex they are.
  4. Estimate the normal distribution of these queries. For example, assuming 50% are simple queries, 30% are complex queries, and 20% are very complex queries.
  5. With all the above, create an estimation of the cost per interaction with your Agent, and add some margin to it.
  6. Map the cost per interaction to credits and the final price of the Pricing Plan.
info

In Nevermined we are working to facilitate the pricing of AI Agents. In the meantime, contact us via Discord if you need any support with this process.