# Answer Generation With MAAX AI Take a look at the image below:

As you can see, whenever you receive a message that you want to be responded to you will be calling MAAX AI API. You will need to perform two separate actions to finally receive a response that can be sent back to your prospect: * Initiate answer generation * Retrieve answer There are multiple reasons our engineering team decided to keep these two seemingly similar functions separate: 1. **More control:** meaning that you can send back the generated message of platform more easily. 2. **Scaling:** We designed MAAX API to handle millions of messages every hour. Keeping those two endpoints separate was pivotal for making possible. 3. **Better integration:** You may want to inegratee the MAAX AI answer generation routines in your existing (custom) tech stack. Keeping these endpoints separate allows your tech team to achieve this. In addition, as you can see there is normally a delay between the answer generation and answer retrieval. Here is why: * **MAAX AI is so much more than a clone of a LLM model:** In the backend MAAX AI performs complex math, data analysis, text analysis and other computationally intense proceedures to craft that perfect response. And it takes some time. Sometimes up to a minute. * **Customer (Prospect) Experience:** the messages we craft are designed to feel as "human" as possible. So if someone sends an email to your customer support and receives an answer within a minute then the prospect is likely to be suspicious of how this was generated. Having a 15 minute delay creates a much more "human" feeling to the entire experience. Thats also why we generally are skeptical about chatbots. {% hint style="info" %} To get a better understanding & intuition about why it makes sense to create some delay between when the sender sends a message and they receive an answer we recommend that you familiarize yourself with a concept of "Uncanny Valley". Here are some great resources to do so: * * {% endhint %}